txt/2109.11406.txt

Named Entity Recognition and Classification on Historical
Documents: A Survey
MAUD EHRMANN, Ecole Polytechnique Fédérale de Lausanne
AHMED HAMDI, University of La Rochelle
ELVYS LINHARES PONTES, University of La Rochelle
MATTEO ROMANELLO, Ecole Polytechnique Fédérale de Lausanne
ANTOINE DOUCET, University of La Rochelle
After decades of massive digitisation, an unprecedented amount of historical documents is available in digital
format, along with their machine-readable texts. While this represents a major step forward with respect
to preservation and accessibility, it also opens up new opportunities in terms of content mining and the
next fundamental challenge is to develop appropriate technologies to efficiently search, retrieve and explore
information from this ‘big data of the past’. Among semantic indexing opportunities, the recognition and
classification of named entities are in great demand among humanities scholars. Yet, named entity recognition
(NER) systems are heavily challenged with diverse, historical and noisy inputs. In this survey, we present the
array of challenges posed by historical documents to NER, inventory existing resources, describe the main
approaches deployed so far, and identify key priorities for future developments.
CCS Concepts: •Computing methodologies →Information extraction ;Machine learning ;Language
resources ;•Information systems →Digital libraries and archives.
Additional Key Words and Phrases: named entity recognition and classification, historical documents, natural
language processing, digital humanities
1 INTRODUCTION
For several decades now, digitisation efforts by cultural heritage institutions are contributing an
increasing amount of facsimiles of historical documents. Initiated in the 1980s with small scale,
in-house projects, the ‘rise of digitisation’ grew further until it reached, already in the early 2000s,
a certain maturity with large-scale, industrial-level digitisation campaigns [ 188]. Billions of images
are being acquired and, when it comes to textual documents, their content is transcribed either
manually via dedicated interfaces, or automatically via optical character recognition (OCR) or
handwritten text recognition (HTR) [ 31,129]. As a result, it is nowadays commonplace for memory
institutions (e.g. libraries, archives, museums) to provide digital repositories that offer rapid, time-
and location-independent access to facsimiles of historical documents as well as, increasingly,
full-text search over some of these collections.
Beyond this great achievement in terms of preservation and accessibility, the availability of
historical records in machine-readable formats bears the potential of new ways to engage with their
contents. In this regard, the application of machine reading to historical documents is potentially
transformative, and the next fundamental challenge is to adapt and develop appropriate technolo-
gies to efficiently search, retrieve and explore information from this ‘big data of the past’ [ 98]. Here
research is stepping up and the interdisciplinary efforts of the digital humanities (DH), natural
language processing (NLP) and computer vision communities are progressively pushing forward
the processing of facsimiles, as well as the extraction, linking and representation of the complex
Authors’ addresses: Maud Ehrmann, [email protected], Ecole Polytechnique Fédérale de Lausanne; Ahmed Hamdi,
[email protected], University of La Rochelle; Elvys Linhares Pontes, [email protected], University of
La Rochelle; Matteo Romanello, [email protected], Ecole Polytechnique Fédérale de Lausanne; Antoine Doucet,
[email protected], University of La Rochelle.arXiv:2109.11406v1  [cs.CL]  23 Sep 20212 Ehrmann et al.
Fig. 1. Swiss journal L’Impartial , issue of 31 Dec 1918. Facsimile of the first page (left), zoom on an article
(middle), and OCR of this article as provided by the Swiss National Library (completed in the 2010s) (right).
information enclosed in transcriptions of digitised collections. In this endeavor, information extrac-
tion techniques, and particularly named entity (NE) processing, can be considered among the first
and most crucial processing steps.
Named entity recognition and classification (NER for short) corresponds to the identification of
entities of interest in texts, generally of the types Person ,Organisation andLocation . Such entities
act as referential anchors which underlie the semantics of texts and guide their interpretation.
Acknowledged some twenty years ago, NE processing has undergone major evolution since then,
from entity recognition and classification to entity disambiguation and linking, and is representative
of the evolution of information extraction from a document- to a semantic-centric view point [ 156].
As for most NLP research areas, recent developments around NE processing are dominated by
deep neural networks and the usage of embedded language representations [ 37,110]. Since their
inception up to now, NE-related tasks are of ever-increasing importance and at the core of virtually
any text mining application.
From the NLP perspective, NE processing is useful first and foremost in information retrieval,
or the activity of retrieving a specific set of documents within a collection given an input query.
Guo et al. [ 78] as well as Lin et al. [ 118] showed that more than 70% of queries against modern
search engines contain a named entity, and it has been suggested that more than 30% of content-
bearing words in news text correspond to proper names [ 69]. Entity-based document indexing
is therefore desirable. NEs are also highly beneficial in information extraction, or the activity of
finding information within large volumes of unstructured texts. The extraction of salient facts about
predefined types of entities in free texts is indeed an essential part of question answering [ 127],
media monitoring [ 182], and opinion mining [ 9]. Besides, NER is helpful in machine translation [ 85],
text summarisation [97], and document clustering [62], especially in a multilingual setting [181].
As for historical material (cf. Figure 1), primary needs also revolve around retrieving documents
and information, and NE processing is of similar importance [ 35]. There are less query logs over
historical collections than for the contemporary web, but several studies demonstrate how prevalent
entity names are in humanities users’ searches: 80% of search queries on the national library of
France’s portal Gallica contain a proper name [ 33], and geographical and person names dominate
the searches of various digital libraries, be they of artworks, domain-specific historical documents,
historical newspapers, or broadcasts [ 14,32,92]. Along the same line, several user studies emphasise
the role of entities in various phases of the information-seeking workflow of historians [ 47,71],
now also reflected in the ‘must-have’ of exploration interfaces, e.g. as search facets over historicalNamed Entity Recognition and Classification on Historical Documents: A Survey 3
newspapers [ 49,145] or as automatic suggestions over large-scale cultural heritage records [ 72].
Besides document indexing, named entity recognition can also benefit downstream processes
(e.g. biography reconstruction [ 64] or event detection [ 176]), as well as various data analysis and
visualisation (e.g. on networks [ 194]). Finally, and perhaps most importantly, NER is the first step of
entity linking, which can support the cross-linking of multilingual and heterogeneous collections
based on authority files and knowledge bases. Overall, entity-based semantic indexing can greatly
support the search and exploration of historical documents, and NER is increasingly being applied
on such a material.
Yet, the recognition and classification of NEs in historical texts is not straightforward, and
performances are rarely on par with what is usually observed on contemporary, well-edited English
news material [ 50]. In particular, NER on historical documents faces the challenges of domain
heterogeneity, input noisiness, dynamics of language, and lack of resources. If some of these issues
have already been tackled in isolation in other contexts (with e.g. user-generated text), what makes
the task particularly difficult is their combination, as well as their magnitude: texts are severely
noisy, domains and time periods are far apart, and there is no (or not yet) historical web to easily
crawl to capture language models. In this context of new material, interests and needs, and in times
of rapid technological change with deep learning, this paper presents a survey of NER research
on historical documents. The objectives are to study the main challenges facing named entity
recognition and classification when applied to historical documents, to inventory the strategies
deployed to deal with them so far, and to identify key priorities for future developments.
Section 2 outlines the objectives, the scope and the methodology of the survey, and Section 3
provides background on NE processing. Next, Section 4 introduces and discusses the challenges of
NER on historical documents. In response, Section 5 proposes an inventory of existing resources,
while Section 6 and 7 present the main approaches, in general and in view of specific challenges,
respectively. Finally, Section 8 discusses next priorities and concludes.
2 FRAMING OF THE SURVEY
2.1 Objectives
This survey focuses on NE recognition and classification, and does not consider entity linking nor
entity relation extraction. With the overall objective of characterising the landscape of NER on
historical documents, the survey reviews the history, the development, and the current state of
related approaches. In particular, we attempt to answer the following questions:
Q1What are the key challenges posed by historical documents to NER?
Q2Which existing resources can be leveraged in this task, and what is their coverage in terms
of historical periods, languages and domains?
Q3Which strategies were developed and successfully applied in response to the challenges faced
by NER on historical documents? Which aspects of NER systems require adaptation in order
to obtain satisfying performances on this material?
While investigating the answers to these questions, the survey will also shed light on the variety of
domains and usages of NE processing in the context of historical documents.
2.2 Document Scope and Methodology
Cultural heritage covers a wide range of material and the document scope of this survey, centred
on ‘historical documents’, needed to be clearly delineated. From a document processing perspective,
there is no specific definition of what a historical document is, but only shared intuitions based on
multiple criteria. Time seems an obvious one, but where to draw the line between historical and
contemporary documents is a tricky question. Other aspects include the digital origin (digitised or4 Ehrmann et al.
Title Type Discipline
Annual Meeting of the Association for Computational Linguistics (ACL) proceedings CL/NLP
Digital Humanities conference proceedings DH
Digital Scholarship in the Humanities (DSH) journal DH
Empirical Methods in Natural Language Processing (EMNLP) proceedings NLP
International Conference on Language Resources and Evaluation (LREC) proceedings NLP
International Journal on Digital Libraries journal DH
Journal of Data Mining and Digital Humanities (JDMDH) journal DH
Journal on Computing and Cultural Heritage (JOCCH) journal DH
Language Resources and Evaluation (LRE) journal NLP
SIGHUM Workshop on Computational Linguistics for Cultural Heritage proceedings CL/NLP/DH
Table 1. Publication venues whose archives were scanned as part of this survey (in alphabetical order).
born-digital), the type of writing (handwritten, typeset or printed), the state of the material (heavily
degraded or not), and of the language (historical or not). None of these criteria define a clear set of
documents and any attempt of definition resorts to, eventually, subjective decisions.
In this survey, we consider as historical document any document of textual nature mainly,
produced or published up to 1979, regardless of its topic, genre, style or acquisition method. The
year 1979 is not arbitrary and corresponds to one of the most recent ‘turning points’ acknowledged
by historians [ 26]. This document scope is rather broad, and the question of the too far-reaching
‘textual nature’ can be raised in relation to documents such as engravings, comics, card boards or
even maps, which can also contain text. In practice, however, NER was mainly applied on printed
documents so far, and these represent most of the material of the work reviewed here.
The compilation of the literature was based on the following strategies: scanning of the archives
of relevant journals and conference series, search engine-based discovery, and citation chaining.
We considered key journals and conference series both in the fields of natural language processing
and digital humanities (see Table 1). For searching, we used a combination of keywords over the
Google Scholar and Semantic Scholar search engines.1With a few exceptions, we only considered
publications that included a formal evaluation.
2.3 Previous surveys and target audience
Previous surveys on NER focused either on approaches in general, giving an overview of features,
algorithms and applications, or on specific domains or languages. In the first group, Nadeau
et al. [ 130] provided the first comprehensive survey after a decade of work on NE processing,
reviewing existing machine learning approaches of that time, as well as typologies and evaluation
metrics. Their survey remained the main reference until the introduction of neural network-based
systems, recently reviewed by Yadav et al. [ 200] and Li et al. [ 116]. The latest NER survey to date
is the one by Nazar et al. [ 131], which focuses specifically on generic domains and on relation
extraction. In the second group, Leaman et al. [ 111] and Campos et al. [ 29] presented a survey
of advances in biomedical named entity recognition, while Lei et al. [ 114] considered the same
domain in Chinese. Shaalan focused on general NER in Arabic [175], and surveys exist for Indian
languages [ 142]. Recently, Georgescu et al. [ 68] focused on NER aspects related to the cybersecurity
domain. Turning our attention to digital humanities, Sporlerder [ 177] and Piotrowski [ 147] provided
general overviews of NLP processing for cultural heritage domains, considering institutional,
documentary and technical aspects. To the best of our knowledge, this is the first survey on the
application of NER to historical documents.
1E.g. ‘named entity recognition’, ‘nerc’, ‘named entity processing’, ‘historical documents’, ‘old documents’ over https:
//scholar.google.com and https://www.semanticscholar.org/Named Entity Recognition and Classification on Historical Documents: A Survey 5
The primary target audiences are researchers and practitioners in the fields of natural language
processing and digital humanities, as well as humanities scholars interested in knowing and
applying NER on historical documents. Since the focus is on adapting NER to historical documents
and not on NER techniques themselves, this study assumes a basic knowledge of NER principles
and techniques; however, it will provide information and guidance as needed. We use the terms
‘historical NER’ and ‘modern NER’ to refer to work and applications which focus on, respectively,
historical and non-historical (as we define them) materials.
3 BACKGROUND
Before delving into NER for historical documents, this section provides a generic introduction to
named entity processing and modern NER (Section 3.1 and 3.2), to the types of resources required
(Section 3.3), and to the main principles underlying NER techniques (Section 3.4).
3.1 NE processing in general
As of today, named entity tasks correspond to text processing steps of increasing level of complexity,
defined as follows:
(1)recognition and classification – or the detection of named entities, i.e. elements in texts
which act as a rigid designator for a referent, and their categorisation according to a set of
predefined semantic categories;
(2)disambiguation/linking – or the linking of named entity mentions to a unique reference in a
knowledge base, and
(3) relation extraction – or the discovery of relations between named entities.
First introduced in 1995 during the 6𝑡ℎMessage Understanding Conference [ 75], the task of NE
recognition and classification (task 1 above) quickly broadened and became more complex, with
the extension and refinement of typologies,2the diversification of languages taken into account,
and the expansion of the linguistic scope with, along proper names, the consideration of pronouns
and nominal phrases as candidate lexical units (especially during the ACE program [ 45]). Later
on, as recognition and classification were reaching satisfying performances, attention shifted to
finer-grained processing, with metonymy recognition [ 123] and fine-grained classification [ 57,122],
and to the next logical step, namely entity resolution or disambiguation (task 2 above, not covered
in this survey). Besides the general domain of clean and well-written news wire texts, NE processing
is also applied to specific domains, particularly bio-medical [ 73,102], and to more noisy inputs such
as speech transcriptions [ 66] and tweets [ 148,159]. In recent years, one of the major developments
of NE processing is its application to historical material.
Importantly, and although the question of the definition of named entities is not under focus here,
we shall specify that we adopt in this regard the position of Nadeau et al. [ 130] for which “ the word
‘Named’ aims to restrict [Named Entities] to only those entities for which one or many rigid designators,
as defined by S. Kripke, stands for the referent ”. Concretely speaking, named entities correspond to
different types of lexical units, mostly proper names and definite descriptions, which, in a given
discourse and application context, autonomously refer to a predefined set of entities of interest.
There is no strict definition of named entities, but only a set of linguistic and application-related
criteria which, eventually, compose a heterogeneous set of units.3
Finally, let us mention two NE-related specific research directions: temporal information process-
ing and geoparsing. This survey does not consider work related to temporal analysis and, when
relevant, occasionally mentions some related to geotagging.
2See e.g. the overviews of Nadeau et al. [130, pp. 3-4] and Ehrmann et al. [51].
3See Ehrmann [48, pp.81-188] for an in-depth discussion of NE definition.6 Ehrmann et al.
Table 2. Illustration of IOB tagging scheme (example 1).
Tokens (X) NER label (Y) POS Chunk
Switzerland B-LOC NNP I-NP
stands O VBZ I-VP
accused O VBN I-VP
by O IN I-PP
Senator O NNP I-NP
Alfonse B-PER NNP I-NP
D’Amato I-PER NNP I-NP
... ... ... ...
3.2 NER in a nutshell
3.2.1 A sequence labelling task. Named entity recognition and classification is defined as a sequence
labelling task where, given a sequence of tokens, a system seeks to assign labels (NE classes) to this
sequence. The objective for a system is to observe, in a set of labelled examples, the word-labels
correspondences and their most distinctive features in order to learn identification and classification
patterns which can then be used to infer labels for new, unseen sequences of tokens. This excerpt
from the CoNLL-03 English test dataset [ 190] illustrates a training example (or the predictions a
system should output):
(1)[𝐿𝑂𝐶 Switzerland] stands accused by Senator [𝑃𝐸𝑅Alfonse D’Amato], chairman of the powerful [𝑂𝑅𝐺
U.S. Senate Banking Committee], of agreeing to give money to [𝐿𝑂𝐶 Poland] (...)
Such input is often represented with the IOB tagging scheme, where each token is marked as being
at the beginning (B), inside (I) or outside (O) of an entity of a certain class [ 155]. Fig. 2 represents
the above example in IOB format, from which systems try to extract features to learn NER models.
3.2.2 Feature space. NER systems’ input corresponds to a linear representation of text as a sequence
of characters, usually processed as a sequence of words and sentences. This input is enriched with
features or ‘clues’ a system consumes in order to learn (or generalise) a model. Typical NER
features may be observed at three levels: words, close context or sentences, and document. At the
morphological level, features include e.g. the word itself, its length, whether it is (all) capitalised or
not, whether it contains specific word patterns or specific affixes (e.g. the suffixes -vitch or-sson
for person names in Russian and Swedish), its base form, its part of speech (POS), and whether
it is present in a predefined list. At the contextual level, features reflect the presence or absence
of surrounding ‘trigger words’ (or combination thereof, e.g. Senator andtopreceding a person or
location name, Committee ending an organisation name), or of surrounding NE labels. Finally, at
the document level, features correspond to e.g. the position of the mention in the document or
paragraph, the occurrence of other entities in the document, or the document metadata.
These features can be absent or ambiguous, and none of them is systematically reliable; it is
therefore necessary to combine them, and this is where statistical models are helpful. Features are
observed in positive and negative examples, and are usually also encoded according to the IOB
scheme (e.g. part-of-speech and chunk annotation columns in Fig. 2). In traditional, feature-based
machine learning, features are specified by the developer (feature engineering), while in deep
learning they are learned by the system itself (feature learning) and go beyond those specified
above.
3.2.3 NER evaluation. Systems are evaluated in terms of precision (P), recall (R) and F-measure
(F-score, the harmonic mean of P and R). Over the years, different scoring procedures and measuresNamed Entity Recognition and Classification on Historical Documents: A Survey 7
were defined in order to take into account various phenomena such as partial match or incorrect
type but correct mention, or to assign different weights to various entity and/or error types. These
fine-grained evaluation metrics allow for a better understanding of the system’s performance
and for tailoring the evaluation to what is relevant for an application. Examples include the
(mostly abandoned) ACE ‘entity detection and recognition value’ (EDR), the slot error rate (SER) or,
increasingly, the exact vs. fuzzy match settings where entity mention boundaries need to correspond
exactly vs. to overlap with the reference. We refer the reader to [ 130, pp.12-15], [ 116, pp.3-4] and
[136, chapter 6]. This survey reports systems’ performances in terms of P, R and F-score.
3.3 NER resource types
Resources are essential when developing NER systems. Four main types of resources may be
distinguished, each playing a specific role.
3.3.1 Typologies. Typologies define a semantic framework for the entities under consideration.
They corresponds to a formalised and structured description of the semantic categories to consider
(the objects of the world which are of interest), along with a definition of their scope (their realisation
in texts). There exist different typologies, which can be multi-purpose or domain-specific, and with
various degrees of hierarchisation. Most of them are defined and published as part of evaluation
campaigns, with no tradition of releasing typologies as such outside this context. Typologies form
the basis of annotation guidelines, which explicit the rules to follow when manually annotating a
corpus and are crucial for the quality of the resulting material.
3.3.2 Lexicons and knowledge bases. Next, lexicons and knowledge bases provide information
about named entities which may be used by systems for the purposes of recognition, classification
and disambiguation. This type of resource has evolved significantly in the last decades, as a result
of the increased complexity of NE-related tasks and of technological progress made in terms of
knowledge representation. Information about named entities can be of lexical nature, relating to the
textual units making up named entities, or of encyclopædic nature, concerning their referents. The
first case corresponds to simple lists named lexica or ‘gazetteers’4which encode entity names, used
in look-up procedures, and trigger words, used as features to guess names in texts. The second case
corresponds to knowledge bases which encode various non-linguistic information about entities
(e.g. date of birth/death, alma mater, title, function), used mainly for entity linking (Wikipediaand
DBpedia [ 113] being amongst the best-known examples). With the advent of neural language
models, the use of explicit lexical information stored in lexica could have been definitely sealed,
however gazetteer information still proves useful when incorporated as feature concatenated to
pre-trained embeddings [37, 89], confirming that NER remains a knowledge-intensive task [157].
3.3.3 Word embeddings and language models. Word embeddings are low-dimensional, dense vec-
tors which represent the meaning of words and are learned from word distribution in running
texts. Stemming from the distributional hypothesis, they are part of the representation learning
paradigm where the objective is to equip machine learning algorithms with generic and efficient
data representations [ 16]. Their key advantage is that they can be learned in a self-supervised fash-
ion, i.e. from unlabelled data, enabling the transition from feature engineering to feature learning.
The principle of learning and using distributional word representations for different tasks was
already present in [ 13,37,193], but it is with the publication of word2vec, a software package which
provided an efficient way to learn word embeddings from large corpora [ 126], that embeddings
started to become a standard component of modern NLP systems, including NER.
4A term initially devoted to toponyms afterwards extended to any NE type.8 Ehrmann et al.
Since then, much effort has been devoted to developing effective means of learning word rep-
resentations, first moving from words to sub-words and characters, and then from words to
words-in-context with neural language models. The first generation of ‘traditional’ embeddings
corresponds to static word embeddings where a single representation is learned for each word
independently of its context (at the type level). Common algorithms for such context-independent
word embeddings include Google word2vec [ 126], Stanford Glove [ 143] and SENNA [ 37]. The
main drawbacks of such embeddings are their poor modelling of ambiguous words (embeddings
are static) and their inability to handle out-of-vocabulary (OOV) words, i.e. words not present
in the training corpus and for which there is no embedding. The usage of character-based word
embeddings, i.e. word representations based on a combination of its character representations, can
help process OOV words and make better use of morphological information. Such representations
can be learned in a word2vec fashion, as with fastText [ 21], or via CNN or RNN-based architectures
(see Section 3.4 for a presentation of types of networks).
However, even enriched with sub-word information, traditional embeddings are still ignorant of
contextual information. This short-coming is addressed by a new generation of approaches which
takes as learning objective language modelling, i.e. the task of computing the probability distribution
of the next word (or character) given the sequence of previous words (or characters) [ 17]. By taking
into account the entire input sequence, such approaches can learn deeper representations which
capture many facets of language, including syntax and semantics, and are valid for various linguistic
contexts (at the token level). They generate powerful language models (LMs) which can be used for
downstream tasks and from which contextual embeddings can be derived. These LMs can be at the
word level (e.g. ELMo [ 144], ULMFiT [ 88], BERT [ 43] and GPT [ 153]), or character-based such as
the contextual string embeddings proposed by Akbik et al. [ 4] (a.k.a flair embeddings). Overall,
alongside static character-based word and word embeddings, character-level and word-level LM
embeddings are pushing the frontiers in NLP and are becoming key elements of NER systems, be it
for contemporary or historical material.
3.3.4 Corpora. Finally, a last type of resource essential for developing NER systems is labelled
documents and, to some extent, unlabelled textual data. Labelled corpora illustrate an objective
and are used either as a learning base or as a point of reference for evaluation purposes. Unlabelled
textual material is necessary to acquire embeddings and language models.
3.4 NER methods
Similarly to other NLP tasks, NER systems are developed according to three standard families of
algorithms, namely rule-based, feature-based (traditional machine learning) and neural-based (deep
learning).
3.4.1 Rule-based approaches. Early NER methods in the mid-1990s were essentially rule-based.
Such approaches rely on rules manually crafted by a developer (or linguist) on the base of regularities
observed in the data. Rules manipulate language as a sequence of symbols and interpret associated
information. Organised in what makes up a grammar, they often rely on a series of linguistic pre-
processing (sentence splitting, tokenization, morpho-syntactic tagging), require external resources
storing language information (e.g. triggers words in gazetteers) and are executed using transducers.
Such systems have the advantages of not requiring training data and of being easily interpretable,
but need time and expertise for their design.
3.4.2 Machine-learning based approaches. Very popular until the late 1990s, rule-based approaches
were superseded by traditional machine learning approaches when large annotated corpora became
available and allowed the machine learning of statistical models in supervised, semi-supervised, andNamed Entity Recognition and Classification on Historical Documents: A Survey 9
later unsupervised fashion. Traditional, feature-based machine learning algorithms learn inductively
from data on the base of manually selected features. In supervised NER, they include support vector
machines [ 94], decision trees [ 185], as well as probabilistic sequence labelling approaches with
generative models such as hidden markov models [ 19] and discriminative ones such as maximum
entropy models [ 15] and linear-chained conditional random fields (CRFs) [ 109]. Thanks to their
capacity to take into account the neighbouring tokens, CRFs proved particularly well-suited for
NER tagging and became the standard for feature-based NER systems.
3.4.3 Deep learning approaches. Finally, latest research on NER is largely (if not exclusively) domi-
nated by deep learning (DL). Deep learning systems correspond to artificial neural networks with
multiple processing layers which learn representations of data with multiple levels of abstrac-
tion [ 112]. In a nutshell, (deep) neural networks are composed of computational units, which take
a vector of input values, multiply it by a weight vector, add a bias, apply a non-linear activation
function, and produce a single output value. Such units are organised in layers which compose
a network, where each layer receives its input from the previous one and passes it to the next
(forward pass), and where parameters that minimise a loss function are learned with gradient
descent (backward pass). The key advantage of neural networks is their capacity to automatically
learn input representations instead of relying on manually elaborated features, and very deep
networks (with many hidden layers) are extremely powerful in this regard.
Deep learning architectures for sequence labelling have undergone rapid change over the last few
years. These developments are function of two decisive aspects for successful deep learning-based
NER: at the architecture level, the capacity of a network to efficiently manage context, and, at the
input representation level, the capacity to benefit from or learn powerful embeddings or language
models. In what follows we briefly review main deep learning architectures for modern NER and
refer the reader to Lin et al. [116] for more details.
Motivated by the desire to avoid task-specific engineering as much as possible, Collobert et al. [ 37]
pioneered the use of neural nets for four standard NLP tasks (including NER) with convolutional
neural networks (CNN) that made used of trained type-level word embeddings and were learned in
an end-to-end fashion. Their unified architecture SENNA5reached very competitive results for
NER ( 89 .86%F-score on the CoNLL-03 English corpus) and near state-of-the-art results for the
other tasks. Following Collobert’s work, developments focused on architectures capable of keeping
information of the whole sequence throughout hidden layers instead of relying on fixed-length
windows. These include recurrent neural networks (RNN), either simple [ 59] or bi-directional [ 170]
(where input is processed from right to left and from left to right), and their more complex variants
of long short-term memory networks (LSTM) [ 86] and gated recurrent units (GRU) [ 34] which
mitigate the loss of distant information often observed in RNN. Huang et al. [ 89] were among the
first to apply a bidirectional LSTM (BiLSTM) network with a CRF decoder to sequence labelling,
obtaining 90 .1%F-score on the NER CoNLL-03 English dataset. Soon, BiLSTM networks became
the de facto standard for context-dependent sequence labelling, giving rise to a body of work
including Lample et al. [ 110], Chiu et al. [ 110], and Ma et al. [ 121] (to name but a few). Besides
making use of bidirectional variants of RNN, these work also experiment with various input
representations, in most cases combining learned character-based representations with pre-trained
word embeddings. Character information has proven useful for inferring information for unseen
words and for learning morphological patterns, as demonstrated by the 91 .2%F-score of Ma et
al. [121] on CoNLL-03, and the systematically better results of Lample et al. [ 110] on the same
dataset when using character information. A more recent study by Taillé et al. [ 186] confirms the
role of sub-word representations for unseen entities.
5‘Semantic/syntactic Extraction using a Neural Network Architecture’.10 Ehrmann et al.
The latest far-reaching innovation in the DL architecture menagerie corresponds to self-attention
networks, or transformers [ 196], a new type of simple networks which eliminates recurrence and
convolutions and are based solely on the attention mechanism. Transformers allow for keeping a
kind of global memory of the previous hidden states where the model can choose what to retrieve
from (attention), and therefore use relevant information from large contexts. They are mostly trained
with a language modelling objective and are typically organised in transformer blocks, which can be
stacked and used as encoders and decoders. Major pre-training transformer architectures include the
Generative Pre-trained Transformer (GPT, a left-to-right architecture) [ 153] and the Bidirectional
Encoder Representation from Transformer (BERT, a bidirectional architecture) [ 43], which achieves
92 .8%NER F-score on CoNLL-03. More recently, Yamada et al. [ 201] proposed an entity-aware
self-attention architecture which achieved 94 .3%F-score on the same dataset. Transformer-based
architectures are the focus of extensive research and many model variants were proposed, of which
Tay et al. [187] propose an overview.
Overall, two points should be noted. First, that beyond the race for the leader board (based on the
fairly clean English CoNLL-03 dataset), pre-trained embeddings and language models play a crucial
role and are becoming a new paradigm in neural NLP and NER (the ‘NLP’s ImageNet moment’ [ 167]).
Second, that powerful language models are also paving the way for transfer learning, a method
particularly useful with low-resource languages and out-of-domain contexts, as is the case with
challenging, historical texts.
4 CHALLENGES
Named entity recognition on historical documents faces four main challenges for which systems
developed on contemporary datasets are often ill-equipped. Those challenges are intrinsic to the
historical setting, like time evolution and types of documents, and endemic to the text acquisition
process, like OCR noise. This translates into a variable and sparse feature space, a situation com-
pounded by the lack of resources. This section successively considers the challenges of document
type and domain variety, noisy input, dynamics of language, and lack of resources.
4.1 The (historical) variety space
First, NER on historical texts corresponds to a wide variety of settings, with documents of different
types (e.g. administrative documents, media archives, literary works, documentation of archival
sites or art collections, correspondences, secondary literature), of different nature (e.g. articles,
letters, declarations, memoirs, wires, reports), and in different languages, which, moreover, spans
different time periods and encompasses various domains and countless topics. The objective here
is not to inventory all historical document types, domains and topics, but to underline the sheer
variety of settings which, borrowing an expression from B. Plank [ 149], compose the ‘variety space’
NLP is confronted with, intensified in the present case by the time dimension.6
Two comments should be made in connection with this variety. First, domain shift is a well-
known issue for NLP systems in general and for modern NER in particular. While B. Plank [ 149]
and J. Einsenstein [ 56] investigated what to do about bad and non-standard (or non-canonical)
language with NLP in general, Augenstein et al. [ 8] studied the ability of modern NER systems
to generalise over a variety of genres, and Taillé et al. [ 186] over unseen mentions. Both studies
demonstrated a NER transfer gap between different text sources and domains, confirming earlier
findings of Vilain et al. [ 197]. While no studies have (yet) been conducted on the generalisation
6Considering there is no common grounds on what constitutes a domain and that the term is overloaded, Plank proposes
the concept of “variety space”, defined as a “ unknown high-dimensional space, whose dimensions contain (fuzzy) aspects such
as language (or dialect), topic or genre, and social factors (age, gender, personality, etc.), amongst others. A domain forms a
region in this space, with some members more prototypical than others ” [149].Named Entity Recognition and Classification on Historical Documents: A Survey 11
capacities of NER systems within the realm of historical documents, there are strong grounds to
believe that systems are equally impacted when switching domain and/or document type.
Second, this (historical) variety space is all the more challenging as the scope of needs and
applications in humanities research is much broader than the one usually addressed in modern
NLP. For sure the variety space does not differ much between today and yesterday’s documents (i.e.
if we were NLP developers living in the 18C we would be more or less confronted with the same
‘amount’ of variety as today), however here the difference lies in the interest for all or part of this
variety: while NLP developments tend to focus on some well-identified and stable domains/sub-
domains (sometimes motivated by commercial opportunities), the (digital) humanities and social
sciences research communities are likely interested in the whole spectrum of document types and
domains. In brief, if the magnitude of the variety space is more or less similar for contemporary and
historical documents, the range of interests and applications in humanities and cultural heritage
requires—almost by design—the consideration of an expansive array of domains and document
types.
4.2 Noisy input
Next, historical NER faces the challenges of noisy input derived from automatic text acquisition
over document facsimiles. Text is acquired via two processes: 1) optical character recognition (OCR)
and handwritten text recognition (HTR), which recognise text characters from images of printed
and handwritten documents respectively, and 2) optical layout recognition (OLR), which identifies,
orders and classifies text regions (e.g. paragraph, column, header). We consider both successively.
4.2.1 Character recognition. The OCR transcription of the newspaper article on the right-hand
side of Figure 1 illustrates a typical, mid-level noise, with words perfectly readable ( la Belgique ),
others illegible ( pu. s >s « _jnces ), and tokenization problems ( n’à’pas ,le’Conseiller ). While this does
not really affect human understanding when reading, the same is not true for machines which
face numerous OOV words. Be it by means of OCR or HTR, text acquisition performances can be
impacted by several factors, including: a) the quality of the material itself, affected by the poor
preservation and/or original state of documents with e.g. ink bleed-through, stains, faint text, and
paper deterioration; b) the quality of the scanning process, with e.g. an inadequate resolution or
imaging process leading to frame or border noise, skew, blur and orientation problems; or c) as per
printed documents and in absence of standardisation, the diversity of typographic conventions
through time including e.g. varying fonts, mixed alphabets but also diverse shorthand, accents
and punctuation. These difficulties naturally challenge character recognition algorithms which are,
what is more, evolving from one OCR campaign to another, usually conducted at different times by
libraries and archives. As a result, not only the transcription quality is below expectations, but the
type of noise present in historical machine-readable corpora is also very heterogeneous.
Several studies investigated the impact of OCR noise on downstream NLP tasks. While Lo-
presti [ 120] demonstrated the detrimental effect of OCR noise propagation through a typical NLP
pipeline on contemporary texts, Van Strien et al. [195] focused on historical material and found a
consistent impact of OCR noise on the six NLP tasks they evaluated. If sentence segmentation and
dependency parsing bear the brunt of low OCR quality, NER is also affected with a significant drop
of F-score between good and poor OCR (from 87%to63%for person entities). Focusing specifically
on entity processing, Hamdi et al. [ 79,80] confronted a BiLSTM-based NER model with OCR outputs
of the same text but of different qualities and observed a 30 percentage point loss in F-score when
the character error rate increased from 7% to 20%. Finally, in order to assess the impact of noisy
entities on NER during the CLEF-HIPE-2020 NE evaluation campaign on historical newspapers12 Ehrmann et al.
(HIPE-2020 for short),7Ehrmann et al. [ 53] evaluated systems’ performances on various entity
noise levels, defined as the length-normalised Levenshtein distance between the OCR surface form
of an entity and its manual transcription. They found remarkable performance differences between
noisy and non-noisy mentions, and that already as little noise as 0.1 severely hurts systems’ abilities
to predict an entity and may halve their performances. To sum up, whether focused on a single
OCR version of text(s) [ 195], on different artificially-generated ones [ 79], or on the noise present in
entities themselves [ 53], these studies clearly demonstrate how challenging OCR noise is for NER
systems.
4.2.2 Layout recognition. Beside incorrect character recognition, textual input quality can also be
affected by faulty layout recognition. Two problems surface here. The first relates to incorrect page
region segmentation which mixes up text segments and produces, even with correct OCR, totally
unsuitable input (e.g. a text line reading across several columns). Progress in OLR algorithms makes
this problem rarer, but it is still present for collections processed more than a decade ago. The
second has to do with the unusual text segmentation resulting from correct OLR of column-based
documents, with very short line segments resulting in numerous hyphenated words (cf. Figure 1).
The absence of proper sentence segmentation and word tokenization also affects performances, as
demonstrated in HIPE-2020, in particular Boros et al [ 25], Ortiz Suárez et al . [137] and Todorov et
al. [191] (see Section 6.3).
Overall, OCR and OLR noises lead to a sparser feature space which greatly affects NER perfor-
mances. What makes this ‘noisiness’ particularly challenging is its wide diversity and range: an
input can be noisy in many different ways, and be little to very noisy. Compared to social media,
for which Baldwin et al. [ 10] demonstrated that there exists a noise similarity from a medium to
another (blog, Twitter, etc.) and that this noise is mostly ‘NLP-tractable’, OCR and OLR noises in
historical documents appear as real moving targets.
4.3 Dynamics of language
Another challenge relates to the effects of time and the dynamics of language. As a matter of fact,
historical languages exhibit a number of differences with modern ones, having an impact on the
performances of NLP tools in general, and of NER in particular [147].
4.3.1 Historical spelling variations. The first source of difficulty relates to spelling variations across
time, due either to the normal course of language evolution or to more prescriptive orthographic
reforms. For instance, the 1740 edition of the dictionary of the French Academy (which had 8
editions between 1694 and 1935) introduced changes in the spelling of about one third of the French
vocabulary and, in Swedish 19C literary texts, the letters <f/w/e/q> were systematically used instead
of <v/v/ä/k> in modern Swedish [ 23]. NER can therefore be affected by poor morpho-syntactic
tagging over such morphological variety, and by spelling variation of trigger words and of proper
names themselves. While the latter are less affected by orthographic reforms, they do vary through
time [23].
4.3.2 Naming conventions. Changes in naming conventions, particularly for person names, can also
be challenging. Let alone the numerous aristocratic and military titles that were used in people’s
addresses, it was, until recently, quite common to refer to a spouse using the name of her husband
(which affects more the linking than recognition), and to use now outdated addresses, e.g. the
French expression sieur . These changes have been studied by Rosset et al. [ 165] who compared the
structure of entity names in historical newspapers vs. in contemporary broadcast news. Differences
7https://impresso.github.io/CLEF-HIPE-2020/Named Entity Recognition and Classification on Historical Documents: A Survey 13
include the prevalence of the structure title + last name vs.first + last name forPerson in historical
newspapers and contemporary broadcast news respectively, and of single-component names vs.
multiple-component names for Organisation (idem). Testing several classifiers, the authors also
showed that it is possible to predict the period of a document from the structure of its entities,
thus confirming the evolution of names over time. For their part, Lin et al. [ 117] studied the
generalisation capacities of a state-of-the-art neural NER system on entities with weak name
regularity in a modern corpus and concluded that name regularity is critical for supervised NER
models to generalise over unseen mentions.
4.3.3 Entity and context drifts. Finally, a further complication comes from the historicity of entities,
also known as entity drift, with places, professions, and types of major entities fading and emerging
over time. For instance, a large part of profession names, which can be used as clues to recognise
persons, has changed from the 19C to the 21C.8This dynamism is still valid today (NEs are an open
class) and its characteristics as well as its impact on performances is particularly well documented
for social media: Fromreide et al. showed a loss of 10 F-score percentage points between two Twitter
corpora sampled two years apart [ 65], and Derczynski et al. systematised the analysis with the
W-NUT2017 shared task on novel and emerging entities where, on training and test sets with very
little entity overlaps, the maximum F-score was only 40%[42]. Besides confirming some degree of
‘artificiality’ of classical NE corpora where the overlap between mentions in the train and the test
sets do not reflect real-life settings, these studies illustrate the poor generalisation capacities of
NER systems to unseen mentions due to time evolution. How big and how quick is entity drift in
historical corpora? We could not find any quantitative study on this, but a high variability of the
global referential frame through time is more than likely.
Overall, the dynamics of language represent a multi-faceted challenge where the disturbing factor
is not anymore an artificially introduced noise like with OCR and OLR, but the naturally occurring
alteration of the signal by the effects of time. Both phenomena result in a sparser feature space,
but the dynamics of language appear less elusive and volatile than OCR. Compared to OCR noise,
its impact on NER performances is however relatively under-studied, and only a few diachronic
evaluations were conducted on historical documents so far. Worth of mention is the evaluation
of several NER systems on historical newspaper corpora spanning ca. 200 years, first with the
study of Ehrmann et al. [ 50], second on the occasion of the HIPE-2020 shared task [ 53]. Testing the
hypothesis of the older the document, the lower the performance, both studies reveal a contrasted
picture with non-linear F-score variations over time. If a clear trend of increasing recall over time
can be observed in [ 50], further research is needed to distinguish and assess the impact of each of
the aforementioned time-related variations.
4.4 Lack of resources
Finally, the three previous challenges are compounded by a fourth one, namely a severe lack of
resources. As mentioned in Section 3.3, the development of NER systems relies on four types of
resources—typologies, lexicons, embeddings and corpora—which are of particular importance for
the adaptation of NER systems to historical documents.
With respect to typologies, the issue at stake is, not surprisingly, their dependence on time
and domain. While mainstream typologies with few ‘universal’ classes (e.g. Person ,Organisation ,
Location , and a few others) can for sure be re-used for historical documents, this obviously does not
mean that they are perfectly suited to the content or application needs of any particular historical
collection. Just as universal entity types cannot be used in all contemporary application contexts,
8See for example the variety of occupations in the HISCO database: iisg.amsterdam/en/data/data-websites/history-of-work14 Ehrmann et al.
neither can they be systematically applied to all historical documents: only a small part can be
reused, and they require adaptation. An example is warships, often mentioned in 19C documents,
for which none of the mainstream typologies has an adequate class. To say that typologies need
to be adapted is almost a truism, but it is worth mentioning for it implies that the application of
off-the-shelf NER tools–as is often done–is unlikely to capture all entities of interest in a specific
collection and, therefore, is likely to penalise subsequent studies.
Besides the (partial) inadequacy of typologies, the lack of annotated corpora severely impedes the
development of NER systems for historical documents, for both training and evaluation purposes.
While unsupervised domain adaptation approaches are gaining interest [ 154], most methods still
depend on labelled data to train their models. Little training data usually results in inferior perfor-
mances, as demonstrated—if proof were needed—by Augenstein et al. for NER on contemporary
data [ 8, p. 71], and by Ehrmann et al. on historical newspapers [ 53, Section 7]. NE-annotated
historical corpora exist, but are still rare and scattered over time and domains (cf. Section 5). This
paucity also affects systems’ evaluation and comparison which, besides the lack of gold standards,
is also characterised by fragmented and non-standardised evaluation approaches. The recently
organised CLEF-HIPE-2020 shared task on NE processing in multilingual and historical newspapers
is a first step towards alleviating this situation [53].
Last but not least, if large quantities of textual data are being produced via digitisation, several
factors slow down their dissemination and usage as base material to acquire embeddings and
language models. First, textual data is acquired via a myriad of OCR softwares which, despite the
definition of standards by libraries and archives, supply quite disparate and heavy-to-process output
formats [ 52,164]. Second, even when digitised, historical collections are not systematically openly
accessible due to copyright restrictions. Despite the recent efforts and the growing awareness of
cultural institutions of the value of such assets for machine learning purposes [ 139], these factors
still hamper the learning of language representations from large amounts of historical texts.
Far from being unique to historical NER, lack of resources is a well-known problem in modern
NER [ 51], and more generally in NLP [ 96]. In the case at hand, the lack of resources is exacerbated
by the somewhat youth of the research field and the relatively low attention towards the creation of
resources compared to other domains. Moreover, considering how wide is the spectrum of domains,
languages, document types and time periods to cover, it is likely that a certain resource sparsity will
always remain. Finding ways to mitigate the impact of the lack of resources on system development
and performances is thus essential.
Conclusion on challenges . NER on historical documents faces four main challenges, namely
historical variety space, noisy input, dynamics of language, and lack of resources. If none of
these challenges is new per se—which does not lessen their difficulty—, what makes the situation
particularly challenging is their combination, in what could somehow be qualified an ‘explosive
cocktail’. This set of challenges has two main characteristics: first, the prevalence of the time
dimension, which not only affects language and OCR quality but also causes domain and entity
drifts; and, second, the intensity of the present difficulties, with OCR noise being a real moving
target, and domains and (historical) languages being highly heterogeneous. As a result, with
feature sparsity adding up to multiple confounding factors, systems’ learning capacities are severely
affected. NER on historical documents can therefore be cast as a domain and time adaptation
problem, where approaches should be robust to non-standard, historical inputs, what is more in
a low-resource setting. A first step towards addressing these challenges is to rely on appropriate
resources, discussed in the next section.Named Entity Recognition and Classification on Historical Documents: A Survey 15
5 RESOURCES FOR HISTORICAL NER
This section surveys existing resources for historical NER, considering typologies and annotation
guidelines, annotated corpora, and language representations (see Section 3.3 for a presentation of
NER resource types). Special attention is devoted to how these resources distribute over languages,
domains and time periods, in order to highlight gaps that future efforts should attempt to fill.
5.1 Typologies and annotation guidelines
Typologies and annotation guidelines for modern NER cover primarily the general and bio-medical
domains, and the most used ones such as MUC [ 76], CoNLL [ 190], and ACE [ 45] consist mainly of
a few high-level classes with the ‘universal’ triad Person ,Organisation andLocation [51]. Although
they are used in various contexts, they do not necessarily cover the needs of historical documents.
To the best of our knowledge, very few typologies and guidelines designed for historical material
were publicly released so far. Exceptions include the Quaero [ 165,166], SoNAR [ 125] and impresso
(used in HIPE-2020) [ 54] typologies and guidelines adapted or developed for historical newspapers
in French, German, and English. Designing guidelines and effectively annotating NEs in historical
documents is not as easy as it sounds and peculiarities of historical texts must be taken into account.
These include for example OCRed text, with the question of how to determine the boundaries
of mentions in gibberish strings, and historical entities, with the existence of various historical
statuses of entities through times (e.g. Germany has 8 Wikidata IDs over the 19C and 20C [ 55,
pp.9-10]).
5.2 Annotated corpora
Annotated corpora correspond to sets of documents manually or semi-automatically tagged with
NEs according to a given typology, and are essential for the development and evaluation of NER
systems (see Section 3.3). This section inventories NE-annotated historical corpora documented
in publications and released under an open license.9Their presentation is organised into three
broad groups (‘news’, ‘literature(s)’ and ‘other’), where they appear in alphabetical order. Unless
otherwise noted, all corpora consist of OCRed documents.
Let us start with some observations on the general picture. We could inventory 17 corpora, whose
salient characteristics are summarised in Table 3. It is worth noting that collecting information
about released corpora is far from easy and that our descriptions are therefore not homogeneous. In
terms of language coverage, the majority of corpora are monolingual, and less than a third include
documents written in two or more languages. Overall, these corpora provide support for eleven
currently spoken languages and two dead languages (Coptic and Latin). With respect to corpus
size, the number of entities appears as the main proxy and we distinguish between small (< 10k),
medium (10-30k), large (30-100k) and very large corpora (> 100k).10In the present inventory, very
large corpora are rather exceptional; roughly one third of them are small-sized, while the remaining
are medium- or large-sized corpora. Next, and not surprisingly, a wide spectrum of domains is
represented, from news to literature. This tendency towards domain specialisation is also reflected
in typologies with, alongside the ubiquitous triad of Person ,Location , and Organisation types, a
long tail of specific types reflecting the information or application needs of particular domains.
Finally, in terms of time periods covered, we observe a high concentration of corpora in the 19C,
directly followed by 20C and 21C, while corpora for previous centuries are either scarce or absent.
9Inventory as of June 2021. The Voices of the Great War corpus [ 27] is not included for not released under an open license.
10For comparison, the CoNLL-03 dataset contains ca. 70k mentions for English and 20k for German [ 190], while OntoNotes
v5.0 contains 194k mentions for English, 130k for Chinese and 34k for Arabic [151].16 Ehrmann et al.
Corpus Doc. type Time period Tag set Lang. # NE s Size License
Quaero Old Press [165] newspapers 19C Quaero fr 147,682 xl elra
Europeana [132] newspapers 19C per,loc,org fr, de, nl 40,801 l cc0
De Gasperi [180] various types 20C per,gpe it 35,491 l cc by-nc-sa
Latin NER [60] literary texts 1C bce-2C per,geo,grp la 7,175 s gpl v3.0
HIMERA [189] medical lit. 19C-21C custom en 8,400 s cc by
Venetian references [36] publications 19C-21C custom Multi 12,879 m cc by
Finnish NER [169] newspapers 19C-20C per,loc,org fi 26,588 m n/a
droc [106] novels 17C-20C custom de 6,013 s cc by
Travel writings [178] travelogues 19C-20C loc en 2,228 sn/a
Czech Hist. NE Corpus [90] newspapers 19C custom cz 4,017 s cc by-nc-sa
LitBank [12] novels 19C-20C ace(w/o wea) en 14,000 l cc by-sa
BIOfid [2] publications 18C-20C extended GermEval de 33,545 l gpl v3.0
HIPE [55] newspapers 18C-21C impresso de, en, fr 19,848 m cc by-nc-sa
BDCamões [74] literary texts 16C-21C custom pt 144,600 xl cc by-nc-nd
Coptic Scriptorium corpora literary texts 3C-5C custom cop 88,068 l cc by
GeoNER [104] literary texts 16C-17C geo fr 264 s lgpl-lr
NewsEye [81] newspapers 19C-20C impresso -comp. de, fr, fi,s v 30,580 l cc by
Table 3. Overview of reviewed NE-annotated historical corpora (ordered by publication year).
5.2.1 News. The first group brings together corpora built from historical newspaper collections.
With corpora in five languages (Czech, Dutch, English, French and German), news emerges as the
best-equipped domain in terms of labelled data availability.
The Czech Historical NE Corpus [ 91] is a small corpus produced out of the year 1872 of the
Czech title Posel od Čerchova . Articles are annotated according to six entity types—persons, institu-
tions, artifacts & objects, geographical names, time expressions and ambiguous entities—which,
despite being custom, bear substantial similarities with major typologies. The corpus was manually
annotated by two annotators with an inter-annotator agreement (IAA) of 0.86 (Cohen’s Kappa).
Europeana NER corpora11[132] is a large-sized collection of NE-annotated historical newspaper
articles in Dutch, French and German, containing primarily 19C materials. These corpora were
sampled from the Europeana newspaper collection [ 133] by randomly selecting 100 pages from all
titles for each language, considering only pages with a minimum word-level accuracy of 80%. Three
entity types were considered (person, location, organisation), yet no IAA for the annotations is
reported. Instead, the quality and usefulness of these annotated corpora were assessed by training
and evaluating the Stanford CRF NER classifier (see Section 3.4.2).
The Finnish NER corpus12[169] is composed of a selection of pages from journals and newspapers
published between 1836 and 1918 and digitized by the national library of Finland. The OCR of this
medium-size corpus was manually corrected by librarians and NE annotations were made manually
for half of them, semi-automatically for the other (via the manual correction of the output of a
Stanford NER system trained on the manually corrected subset). Overall, the annotations show a
good IAA of 0.8 (Cohen’s kappa).
The HIPE corpus13[55] is a medium-sized, historical news corpus in French, German and English,
created as part of HIPE-2020. It consists of newspaper articles sampled from Swiss, Luxembourgish
and American newspaper collections covering a time span of ca. 200 years (1798-2018). OCR
quality of the corpus corresponds to real-life setting and varies depending on the digitisation time
and preservation state of original documents. The corpus was annotated following the impresso
11https://github.com/EuropeanaNewspapers/ner-corpora
12https://digi.kansalliskirjasto.fi/opendata/submit (Digitalia (2017-2019) package).
13Version 1.3, https://github.com/impresso/CLEF-HIPE-2020/tree/master/dataNamed Entity Recognition and Classification on Historical Documents: A Survey 17
guidelines [ 54], which are based on and are retro-compatible with the Quaero guidelines [ 166].
The annotation tag set comprises 5 coarse-grained and 23 fine-grained entity types, and includes
entity components as well as nested entities. Wrongly OCRed entity surface forms are manually
corrected and entities are linked towards Wikidata. NERC and EL annotations reached an average
IAA across languages of 0.8 (Krippendorf’s alpha).
The NewsEye dataset14[81] is a large-sized corpus composed of articles extracted from news-
papers published between mid 19C and mid 20C in French, German, Finnish, and Swedish. Four
entity types were considered (person, location, organisation and human product) and annotated
according to guidelines15similar to the impresso ones; entities are linked towards Wikidata and
articles are further annotated with authors’ stances. The annotation reaches high IAAs exceeding
0.8 for Swedish and 0.9 for German, French and Swedish (Cohen’s kappa).
The Quaero Old Press Extended NE corpus16[165] is a very large annotated corpus composed of
295 pages sampled from French newspapers of December 1890. The OCR quality is rather good, with
a character and word error rates of 5% and 36.5% respectively. Annotators were asked to transcribe
wrongly OCRed entity surface forms—similarly to what was done for the HIPE corpus—which
makes both corpora suitable to check the robustness of NER systems to OCR noise. The annotator
agreement on this corpus reaches 0.82 (Cohen’s Kappa).
5.2.2 Literature(s). The second group of corpora relates to literature and is more heterogeneous in
terms of domains and document types, ranging from literary texts to scholarly publications.
To begin with, two resources consist of ancient literary texts. First, the Latin NER corpus17[60]
comprises ancient literary material sampled from three texts representatives of different literary
genres (prose, letters and elegiac poetry) and spanning over three centuries. The annotation tag set
covers persons, geographical place names and group names (e.g. ‘Haeduos’, a Gallic tribe). Next,
the Coptic Scriptorium corpus18is a large-sized collection of literary works written in Coptic, the
language of Hellenistic era Egypt (3C-5C CE), and belonging to multiple genres (hagiographic
texts, letters, sermons, martyrdoms and the Bible). Besides lemma and POS tags, this corpus also
contains (named and non-named) entity annotations, with links towards Wikipedia. In addition to
persons, places and organisations, the entity types include abstract entities (e.g. ‘humility’), animals,
events, objects (e.g. ‘bottles’), substances (e.g. ‘water’) and time expressions. Entity annotations
were produced automatically (resulting in 11k named entities and 6k linked entities), a subset of
which was manually corrected (2,4k named entities and 1,5k linked entities).
Then, several corpora were designed to support computational literary analysis. This is the case
of the BDCamões Collection of Portuguese Literary Documents19[74], a very large annotated
corpus composed of 208 OCRized texts (4 million words) representative of 14 literary genres and
covering five centuries of Portuguese literature (16C-21C). Due to the large time span covered, texts
adhere to different orthographic conventions. Named entity annotations correspond to locations,
organisations, works, events and miscellaneous entities, and were automatically produced (silver
annotations). They constitute only one of the many layers of linguistic annotations of this corpus,
alongside POS tags, syntactic analysis and semantic roles. Next, the LitBank20[12] dataset is a
medium-sized corpus composed of 100 English literary texts published between mid 19C and
beginning 20C. Entities were annotated following the ACE guidelines—with the only exception
14Version 1.0, https://doi.org/10.5281/zenodo.4573313
15https://zenodo.org/record/4574199
16http://catalog.elra.info/en-us/repository/browse/ELRA-W0073/
17https://github.com/alexerdmann/Herodotos-Project-Latin-NER-Tagger-Annotation
18https://github.com/copticscriptorium/corpora
19https://portulanclarin.net/
20https://github.com/dbamman/litbank18 Ehrmann et al.
of weapons as rarely attested—and include noun phrases as well as nested entities. Finally, the
Deutsches ROman Corpus (DROC) [ 106] is a set of 90 richly-annotated fragments of German novels
published between 1650 and 1950. The DROC corpus is enriched with character mentions, character
co-references, and direct speech occurrences. It features more than 50,000 character mentions, of
which only 12% (6,013) contain proper names and thus correspond to traditional person entity
mentions (others correspond to pronouns or appellatives).
Next, two of the surveyed corpora in this group focus specifically on place names. First, Travel
writings21[178] is a small corpus of 38 English travelogues printed between 1850 and 1940. Its tag
set consists of a single type ( Location ), which encompasses geographical, political and functional
locations, thus corresponding to ACE’s gpe,locandfacentity types altogether. Second, the
GeoNER corpus22[104] is a very small corpus consisting of three 16C-17C French literary texts by
Racine, Molière and Marguerite de Valois. Each annotated text is available in its original version, as
well as with automatic and manual historical spelling normalization. Despite its limited size, this
corpus can be a valuable resource for researchers investigating the effects of historical normalisation
on NER.
Finally, moving from literature to scholarly literature, three corpora should be mentioned. First,
BIOfid23[2] is a large NE-annotated corpus composed of ca. 1000 articles sampled from German
books and scholarly journals in the domain of biodiversity and published between 18C and 20C. The
annotation guidelines used for this corpus build upon those used for the GermEval dataset [ 18], with
the addition of time expressions and taxonomies ( Taxon ), i.e. systematic classifications of organisms
by their characteristics (e.g. “northern giant mouse lemur”). Second, HIstory of Medicine CoRpus
Annotation (HIMERA)24[189] is a small-sized corpus in the domain of medical history, consisting
of journal articles and medical reports published between 1840 and 2013. This corpus is annotated
with NEs according to a custom typology comprising, for example, medical conditions, symptoms,
or biological entities. While all annotations were performed on manually corrected OCR output, the
annotation of certain types was carried out in a semi-automatic fashion. Globally, the annotation
reaches good IAAs of 0.8 and 0.86 for exact and relaxed match respectively (F-score). Third, the
Venetian References corpus25[36] contains about 40,000 annotated bibliographic references from a
corpus of books and journal articles on the history of Venice (19C-21C century) in Italian, English,
French, German, Spanish and Latin. Components of references (e.g. author, title, publication date,
etc.) are annotated according to a custom tag set of 26 tags, and references themselves are classified
according to the type of work they refer to (e.g. primary vs. secondary sources).
5.2.3 Other. We found one corpus in the domain of political writings. The De Gasperi corpus26[192]
consists of the complete collection of public documents by Alcide De Gasperi, Italy’s Prime Minister
in office from 1945 to 1953 and one of the founding fathers of the European Union. This large
corpus includes 2,762 documents published between 1901 and 1954 and belonging to a wide variety
of genres. It was automatically annotated with parts of speech, lemmas, person and place names (by
means of TextPro [ 146]). This corpus consists of clean texts extracted from the electronic versions
of previously published volumes.
21https://github.com/dhfbk/Detection-of-place-names-in-historical-travel-writings
22https://github.com/PhilippeGambette/GeoNER-corpus
23https://github.com/FID-Biodiversity/BIOfid
24http://www.nactem.ac.uk/himera/
25https://github.com/dhlab-epfl/LinkedBooksReferenceParsing
26https://github.com/StefanoMenini/De-Gasperi-s-CorpusNamed Entity Recognition and Classification on Historical Documents: A Survey 19
5.3 Language representations
As distributional representations, embeddings and language models need to be trained on large
textual corpora in order to be effective. There exist several large-scale, diachronic collections of
historical documents, such as the Europeana Newspaper collection [ 133], the Trove Newspaper
corpus [ 30], the Digi corpus [ 99], and the impresso public corpus [ 52] (to mention but a few), which
are now used to acquire historical language representations. Given their usefulness in many NLP
tasks, embeddings and language models are increasingly shared by researchers, thus constituting a
growing and quickly evolving pool of resources that can be used in historical NER. This section
inventories existing historical language representations, an overview of which is given in Table 4.
5.3.1 Static embeddings. As to traditional word embeddings, we could inventory two main re-
sources. Sprugnoli et al. [ 179] have released a collection of pre-trained word and sub-word English
embeddings learned from a subset of the Corpus of Historical American English [ 40], considering
37k texts published between 1860 and 1939 amounting to about 198 million words. These embed-
dings of 300 dimensions are available according to three types of word representations: embeddings
based on linear bag-of-words contexts (GloVe [ 143]), on dependency parse-trees (Levy et al. [ 115]),
and on bag of character n-grams (fastText [ 21]).27Doughman et al. Doughman et al . [46] have
created Arabic word embeddings from three Lebanese news archives, with materials published
between 1933 and 2011.28Archive-level as well as decade-level embeddings were trained using
word2vec with a continuous bag of words model. Given the imperfect OCRed, hyper-parameter
tuning was used to maximise accuracy on a set of analogy tasks.
Another set of traditional word embeddings consists of diachronic or dynamic embeddings, i.e.
static embeddings trained on different time bins of a corpus and thereafter aligned according to
different strategies (post-hoc alignment after training on different time bins, or incremental training).
Such resources provide a view of words over time and are usually used in diachronic studies such
as culturomics and semantic change, but can also be used to feed neural architectures for other
tasks. Some of the pioneers in releasing such material were Hamilton et al. [ 82], who published a
collection of diachronic word embeddings29for English, French, German and Chinese, covering
roughly 19C-20C. They were computed from many different corpora by using word2vec skip-gram
with negative sampling. Later on, Hengchen et al. [ 83] released a set of diachronic embeddings
of the same type in English, Dutch, Finnish and Swedish trained on large corpora of 19C-20C
newspapers.30More recently, Hengchen et al. [ 84] pursued these efforts with the publication of
diachronic word2vec and fastText models trained on a large corpus of Swedish OCRed newspapers
(1645-1926) (the Kubhist 2 corpus, 5.5 billion tokens). Thanks to its ability to capture sub-word
information, their fastText model allows for retrieving OCR misspellings and spelling variations,
thus being a useful resource for post-OCR correction and historical normalisation.
5.3.2 Contextualised embeddings. Historical character-level LM embeddings are currently avail-
able for German, French, and English. For historical German, Schweter et al. [ 172] have trained
contextualised string embeddings (flair) on articles from two titles from the Europeana newspaper
collection, the Hamburger Anzeiger (about 741 million tokens, 1888-1945) and the Wiener Zeitung
(some 801 million tokens, 1703-1875). Resulting embeddings are part of the Flair library.31Next,
in the context of the HIPE-2020 shared task, fastText word embeddings and flair contextualised
27For the link to the published embeddings see https://github.com/dhfbk/Histo.
28Models as well as evaluation details can be found at: https://doi.org/10.5281/zenodo.3538880.
29https://nlp.stanford.edu/projects/histwords/
30https://zenodo.org/record/3270648
31With the ID de-historic-ha-X (HHA) and de-historic-wz-X (WZ) respectively.20 Ehrmann et al.
Publication Type(s) Model(s) Language(s) Training Corpus
Hamilton et al. [82] classic word embeddings PPMI, SVD, word2vec de, fr, en, cn Google Books +COHA
Hengchen et al. [83] classic word embeddings word2vec en, nl, fi, se newspapers and periodicals
Hengchen et al. [84] char.-based word & word embeddings fastText, word2vec sv Kubhist 2
Sprugnoli et al. [179] char.-based word & word embeddings dependency-based, fastText, GloVe en CHAE
Doughman et al. [46] classic word embeddings word2vec ar Lebanese News Archives
Ehrmann et al. [52, 55] char.-based word & char.-level LM embeddings fastText, flair de,fr,en impresso corpus
Hosseini et al. [87] all types word2vec, fastText, flair, BERT en Microsoft British Library corpus
Schweter et al. [172] character-level LM embeddings BERT, ELECTRA de, fr Europeana Newspaper corpus
Bamman et al. [11] word-level LM embeddings BERT la various Latin corpora
Table 4. Overview of available word embeddings and LMs trained on historical corpora.
string embeddings were made available as auxiliary resources for participants.32They were trained
on newspaper materials in French, German and English, and cover roughly 18C-21C (full details
in [55] and [ 52]). Similarly, Hosseini et al . [87] published a collection of static (word2vec, fastText)
and contextualised embeddings (flair) trained on the Microsoft British Library (MBL) corpus. MBL
is a large-scale corpus composed of 47,685 OCRed books in English (1760-1900) which cover a
wide range of subject areas including philosophy, history, poetry and literature, for a total of
approximately 5.1 billion tokens. For each architecture, authors released models trained either on
the whole corpus or on books published before 1850.
Word-level LM embeddings trained on historical data are available for Latin, French, German
and English. Latin BERT is a LM for Latin trained on 640 million tokens spanning 22 centuries.33
In order to reach a sufficiently large volume of training material, a wide variety of datasets was
employed including the Perseus Digital Library, the Latin Wikipedia (Vicipaedia), and Latin texts
of the Internet Archive. Extrinsic evaluation of the model was performed on POS tagging and word
sense disambiguation, for which Latin BERT demonstrated state-of-the-art results. For historical
German and French, Schweter [171] published BERT and ELECTRA models trained on two subsets
of the Europeana newspapers corpus, consisting of 8 and 11 billion tokens for German and French
respectively. The German models were evaluated on two historical NE datasets, on which the
ELECTRA models over-performed the BERT ones, leading to an overall improvement on the current
state-of-the-art results reported by Schweter and Baiter [172] . Finally, for 19C English, BERT-based
language models trained on the MBL corpus are available in the histLM model collection [ 87]. One
model was trained on the entire corpus, and additional models were created for different time
slices to enable the study of linguistic and cultural changes over the 19C, by fine-tuning an existing
contemporary model (BERT base uncased).
Conclusion on Resources. Resources for historical NER are not numerous but do exist. A few
typologies and guidelines adapted for historical OCRed texts were published. More and more
annotated corpora are being released, but the 17 that we could inventory here are far from the
121 inventoried in [ 51] for modern NE processing. They are to a large extent built from historical
newspaper collections, a type of document massively digitised during the last years. If historical
newspaper contents lend themselves particularly well to NER, this preponderance could also be
taken as an early warning of the risk of reproducing the news bias already observed for contempo-
rary NLP [ 149]. Besides, NE-annotated historical corpora show a modest degree of multilingualism,
and most of them are published under open licenses. As per language representations, historical
embeddings and language models are not numerous but multiply rapidly.
32Available at files.ifi.uzh.ch/impresso/clef-hipe-2020/ and on Zenodo platform under DOI 10.5281/zenodo.3706808; Flair
embeddings were also integrated into the Flair framework: https://github.com/flairNLP/flair. CC BY-NC 4.0 license applies.
33https://github.com/dbamman/latin-bertNamed Entity Recognition and Classification on Historical Documents: A Survey 21
6 APPROACHES TO HISTORICAL NER
This section provides an overview of existing work on NER for historical documents, organised by
type of approach: rule-based, traditional machine learning and deep learning. The emphasis here
is more on the implementation and settings of historical NER methods, while strategies to deal
with specific challenges—regardless of the method—are presented in Section 7. Since research was
almost exclusively done in the context of individual projects, and since there was no shared gold
standard up to recently, system performances are often not comparable. We therefore report results
only when computed on sufficiently large data and explicitly state when results are comparable.
All works deal with OCRed material unless mentioned otherwise. In absence of obvious thematic
or technical grouping criteria, they are presented in order of publication (oldest to newest). Table 5
presents a synthetic view of the reviewed literature.
6.1 Rule-based approaches
As for modern NER, first NER works dealing with historical documents were mainly symbolic.
Rule-based systems do not require training data and are easily interpretable, but need time and
expertise for designing the rules. Numerous rule-based systems have been developed for modern
NER, and they usually obtain good results on well-formed texts (see Section 3.4.1).
Early work performed NER over historical collections using the GATE language technology
environment [ 38], which supports the manual creation of rules and gazetteers. Those work do
not include formal evaluations but are worth mentioning as early exploration efforts, e.g. the
adaptation of rules and gazetteers by Bontcheva et al. [ 22] to recognise Person ,Location ,Occupation
and Status entity types in 18C English court trials. Among other difficulties, authors mention
historical occupation names not present in gazetteers, orthographic variations (punctuation, spelling,
capitalisation), and person name abbreviations.
Thereafter, most systems relied on custom rule sets and made substantial use of gazetteers, with
the objective of addressing the domain and language peculiarities of historical documents. Jones et
al. [95] designed a rule-based system to extract named entities from the Civil War years (1861-1865)
of the American newspaper the Richmond Times Dispatch (on manually segmented and transcribed
issues). They focus on 10 entity types, some of them specific to the period and the material at
hand such as warships, military units and regiments. Their system consists of three main phases:
gazetteer lookup to extract easily identifiable entities; application of high precision rules to guess
new names; and learning of frequency-based rules (e.g. how often Washington appears as a person
rather than a place, and in which context). Best results are obtained for Location andDate, while
the identification of Person ,Organisation andNewspaper titles is lower. Based on a thorough error
analysis, authors conclude that shorter but historically relevant gazetteers may be better than long
ones, and make a plea for the development of comprehensive domain-specific knowledge resources.
Working on Swedish literary classics from the 19C, Borin et al. [ 23] designed a system made
of multiple modules: a gazetteer lookup and finite-state grammars module to recognise entities, a
name similarity module to address lexical variation, and a document centred module to propagate
labels based on documents’ global evidence. They focused on 8 entity types and evaluated system
modules’ performances on an incremental basis. On all types together, the best F-measure reaches
89%, and recall is systematically lower than precision in all evaluation iterations (evaluation setting
is partial match). The main sources of error are spelling variations, unknown names, and noisy
word segmentation due to hyphenation in the original document.
Grover et al. [ 77] focused on two subsets of the Journal of the House of Lords of Great Britain,
one from the late 17C and the other from early 19C, OCRed with different systems and at different
times. OCR quality is erratic, and suffers from numerous quotation marks as well as from the22 Ehrmann et al.
presence of marginalia and of text portions in Latin. An in-house rule-based system, consisting of a
set of rules applied incrementally with access to a variety of lexica, is applied to recognise person
and place names. Before NE tagging, the system attempts to identify marginalia words and noisy
characters in order to ignore them during parsing. The overall performance is evaluated against
test sets of each period, which comprise significantly more person than location names. Results are
comparable for person names for both 17C and 19C sets (ca. 75%F-score), but the earliest period has
significantly worse performance for locations ( 24 .1%and 66 .5%). In most configurations, precision
is slightly above recall (evaluation setting not specified, most likely exact match). An error analysis
revealed that character misspellings and segmentation errors (broken NEs) were the main factors
impacting performances.
The experiments conducted by Broux et al. [ 28] are part of an initiative aiming at improving
access to texts from the ancient world. Working with a large collection of documentary texts
produced between 800 BCE and 800 CE, including all languages and scripts written on any surface
(mainly papyrological and epigraphical resources), one of the objective is to develop and curate
onomastic lists and prosopographies of non-royal individuals attested as living during this period.34
Authors apply a rule-based system benefiting from a huge onomastic gazetteer covering names,
name variants and morphological variants in several ancient languages and scripts. Rules encode
various sets of onomastic patterns specific to Greek, Latin and Egyptian (Greek names are ‘simpler’
than the often multiple Roman names, e.g. Gaius Iulius Caesar ) and specifically designed to capture
genealogical information. This system is used to speed up manual NE annotation of texts, which
in turn is used for network analysis in order to assist the creation of prosopographies. No formal
evaluation is provided.
Fast-forwarding to contemporary times, Kettunen et al. [ 100] experimented with NER on a
collection of Finnish historical newspapers from late 19C - early 20C. Authors insist on the overall
poor quality of the OCR (word level correctness around 70%−75%), as well as on the fact that
they use an existing rule-based system designed for modern Finnish with no adaptation. Not
surprisingly, this combination leads to rather low results with F-scores ranging from 30%to45%
for the 8 targeted entity types (evaluation setting is exact match). The main sources of errors are
bad OCR and multi-word entities.
A recent work by Platas et al. [ 150] focuses on a set of manually transcribed Medieval Spanish texts
(12C-15C) covering various genres such as legal documents, epic poetry, narrative, or drama. Based
on the needs of literary scholars and historians, the authors defined a custom entity typology of 8
main types (plus sub-types). It covers traditional but also more specific types for the identification
of name parts, especially relevant for Medieval Spanish person names featuring many attributes
and complex syntactic structures ( Don Alfonso por la gracia de Dios rey de Castiella de Toledo de
Leon de Gallizia de Seuilla de Cordoua de Murcia e de Jaen ). The system is composed of several
modules dedicated to recognising names using rules and/or gazetteers, increasing the coverage
using variant generation and matching, and recognising person attributes using dependency parsing.
Evaluated on a manually annotated corpus representative of the time periods and genres of the
collection, the system reached satisfactory results with an overall F-score of 77%, ranging from
74%to87%depending on the entity type (evaluation setting is exact match). As usual, recall is
lower than precision, but differences are not high. Although these numbers are lower than what
neural-based systems can achieve, this demonstrates the capacities and suitability of a carefully
designed rule-based system.
34Onomastic relates to the study of the history and origin of proper names (Oxford English dictionary), and prosopography
relates to the collection and study of information about a person.Named Entity Recognition and Classification on Historical Documents: A Survey 23
Finally, it is also worth mentioning a series of work on the geoparsing of historical and literary
texts. With the aim of analysing the interplay between geographical and fictional landscapes,
Moncla et al . [128] experimented with a rule-based system relying on extensive gazetteers to
recognise names of streets, houses, bridges, etc. in French Parisian novels from the 19C. With
spatial entities featuring a high degree of regularity, the system reached very good results on a
relatively small test set (evaluation settings are not entirely clear). Adapting the existing Edinburgh
Geoparser system (derived from Grover et al . [77] above) for historical texts, Alex et al. [ 5] carried
out experiments to recognise place names in different types of 19C British historical documents.
Besides the impact of OCR errors, main observations are that it is essential to perform place and
person name recognition in tandem in order to better handle homonyms—even when dealing with
place names only—, and that gazetteers need substantial adaptation, with careful switching on
and off of standard vs domain-specific lexica. This system was also applied on a set of historical
Edinburgh-specific documents, this time targeting fine-grained location names and considering
three types of material: OCRed documents from 19C British novels, manually crowd-corrected
OCRed texts from the Project Gutenberg collection, and contemporary (born-digital) texts from
Scottish authors [ 7]. Not surprisingly, place name recognition performs best on contemporary texts
(but remains low with an F-score of 75%), worst on historical OCRed text (F-score 68%), and roughly
in-between on crowd-corrected OCRed documents (F-score 72%). Precision scores are similar across
the three collections, but recall scores vary considerably. Much research has been done on the
geoparsing of cultural heritage material but is not further surveyed here.
Conclusion on rule-based approaches . Symbolic approaches were applied on a large variety
of document types, domains and time periods (see Table 5 for an overview of characteristics).
In general, rule-based systems are modular and almost systematically include gazetteer lookup,
rule incremental application, and variant matching. They have difficulties dealing with noisy and
historical input, for which they require normalisation rules and additional linguistic knowledge.
The number of work we could inventory, from the beginning of the 2000s until today, confirms
the long-standing need for NER on historical documents as well as the suitability of symbolic
approaches that can be better dealt with by non experts. Research nevertheless moved away from
such systems in favour of machine learning ones.
6.2 Traditional Machine Learning Approaches
Machine learning algorithms inductively learn statistical models from annotated data on the basis of
manually selected features (see Section 3.4.2). Heavily researched and applied in the 2000s, machine
learning-based approaches contributed strong baselines for mainstream NER, and were rapidly
adopted for NER on historical documents. In this section we review the usage of such traditional,
pre-neural machine learning approaches on historical material, first considering works which apply
already existing models, second which train new ones.
6.2.1 Applying existing models. Early achievements adopted the ‘off-the-shelf’ strategy with the
application of pre-trained NER systems or web services to various historical documents, mainly
with the objectives of assessing baselines and/or comparing system performances. This is the
case of Rodriquez et al. [ 163], who compared the performances of four NER systems (Stanford
CRF classifier, OpenNLP, AlchemyAPI, and OpenCalais) on two English datasets related to WWII:
individual Holocaust survivor testimonies from the Wiener Library of London and letters of soldiers
from King’s College archive. Evaluated on a small dataset, the recognition of Person ,Location and
Organization reached an F-score between 47%and 54%for the testimonies (Stanford CRF being the
most accurate), and between 32%and 36%for the letters (OpenCalais performing best). Surprisingly,
running the same evaluation on manually corrected OCR did not improve results significantly.24 Ehrmann et al.
Major sources of errors were different ways of naming and metonymy phenomena (e.g. warships
named after people), and lack of background knowledge, especially for organisations.
Along the same line, Ehrmann et al. [ 50] conducted experiments on French historical newspapers
on a diachronic basis (covering 200 years) for the types Person andLocation , with the objective of
investigating whether NER performance degrades when going back in time. Their study includes
four systems representative of major approaches for NER: a rule-based system, a supervised machine
learning one (MaxEnt classifier), and two proprietary web services offering NER functionalities
(AlchemyAPI and DandelionAPI). They showed that, compared to a baseline on contemporary news,
all systems feature degraded performances, both in absolute terms and over time (maximum of 67 .6%
F-score for person names for the best system, with exact match). As for time-based observation,
precision is quite irregular, with several ups and downs for all systems for both entity types, but
recall shows less variability and a slight but regular increase for Person , suggesting that person
names are less stable than location names and therefore better recognised when more recent.
Focusing on the impact of historical language normalisation (in this respect see also Section
7.2), Kogkitsidou et al. [ 104] also used and benchmarked several systems (rule-based and machine
learning) for the recognition of Location names in French literary texts from the 16C and 17C.
When applied without any adaptation, systems features very diverse performances, from very low
(36%) to reasonable ( 70%) F-scores, with rule-based ones being better at precision, and machine
learning ones at recall.
Ritze et al. [ 160] worked on historical records of the English High Court of Admiralty of the
17C and used the Stanford CRF classifier with its default English model to recognise Person and
Location types (others were considered but not evaluated). Given the very specific domain of this
corpus, obtained results were reasonable, with a precision in the 77%for both types (recall was not
reported).
Finally, some adopt the approach of ensembling systems, i.e. of considering NE predictions not
from one but several recognisers, according to various voting strategies. Packer et al. [ 138] applied
three algorithms (dictionary-based, regular expressions-based, and HMM-based) in isolation and in
combination for the recognition of person names in various types of English OCRed documents.
They observed increased performances (particularly a better P/R balance) with a majority vote
ensembling. Won et al. [ 198] worked on British personal archives from 16C and 17C and applied
five different systems to recognise place names. They too observed that the combination of multiple
systems through a majority vote (with a minimum of two to a maximum of three votes) was able to
consistently outperform the individual NER systems.
Mere application of existing systems, these work illustrate the inadequacy of already trained
NER models for historical texts. Performances (and settings) of these baseline studies are extremely
diverse, but the following constants are observed: recall is always the most affected, and the Location
type is usually the most robust.
6.2.2 Training models. Other work trained NER systems anew on custom material. Early attempts
include the experiments of Nissim et al. [ 135] on Location entity type in manually transcribed
Scottish parish registers of the late 18C and early 19C. They trained a maximum entropy tagger
with its in-built standard features on a dataset of ca. 6000 location mentions and obtained very
satisfying performance ( 94 .2%F-score), which they explained by the custom training data and the
binary classification task (location vs non-location).
Subsequently, the most frequently used system is the Stanford CRF classifier35[63], particularly
on historical newspapers. Working with the press collection of the National Library of Australia,
Kim et al. [ 103] evaluated two Stanford CRF models, the default English one trained on CoNLL-03
35https://nlp.stanford.edu/software/CRF-NER.htmlNamed Entity Recognition and Classification on Historical Documents: A Survey 25
Publication Domain Document type Time period Language(s) System Comp.
Rule-based
Bontcheva et al. [22] legal court trials 18C en-GB rule-based
Jones et al. [95] news newspapers mid 19C en-US rule-based
Borin et al. [23] literature literary classics 19C sv rule-based
Grover et al. [77] state parliamentary proc. 17C & 19C en-GB rule-based
Broux and Depauw [28] state papyri 4C-1C bce egy, el, la lookup
Kettunen et al. [100] news newspapers 19C-20C fi rule-based
Alex et al. [5] state/literature parl. proc./classics var en-scotland lookup
Alex et al. [7] literature novels 19C en-scotland lookup
Moncla et al. [128] literature novels 19C fr lookup
Platas et al. [150] literature poetry, drama 12C-15C es rule-based
Traditional machine learning
Nissim et al. [135] admin parish registers 18C-19C en-scotland MaxEnt
Packer et al. [138] mix various - en ensemble
Rodriquez et al. [163] egodocs letters & testimonies WWII en-GB several
Galibert et al. [67] news newspapers 19C fr several
Dinarelli et al. [44] news newspapers 19C fr CRF+PCFG
Ritze et al. [160] state admiralty court rec. 17C en-GB CRF
Neudecker et al. [134] news newspapers 19C-20C de, fr, nl CRF
Passaro et al. [141] state war bulletins 20C it CRF
Kim et al. [103] news newspapers - en CRF
Ehrmann et al. [50] news newspapers 19C-20C fr several
Aguilar et al. [1] news medieval charters 10C-13C la CRF
Erdmann et al. [60] literature classical texts 1C bce-2C la CRF
Ruokolainen et al. [169] news newspapers 19C-20C fi CRF+gaz
Won et al. [198] egodocs letters 17-18C en-GB ensemble
El Vaigh et al. [58] news newspapers ( hipe) 19C-20C de, en, fr CRF
Kogkitsidou et al. [104] literature theatre and memoirs 16C-17C French several
Deep Learning
Riedl et al. [158] news newspapers 19C-20C de BiLSTM-CRF ♢
Rodrigues A. et al. [162] bibliometry journals & monographs 19C-20C multi BiLSTM-CRF
Sprugnoli [178] literature travel writing 19C-20C en-US BiLSTM-CRF
Ahmed et al. [2] biodiversity scholarly pub. 19C-20C de BiLSTM-CRF
Kew et al. [101] literature alpine texts 19C-20C multi BiLSTM-CRF
Schweter et al. [172] news newspapers 19C-20C de BiLSTM-CRF ♢
Labusch et al. [108] news newspapers 19C-20C de BERT ♢
Dekhili and Sadat [41] news newspapers ( hipe) 19C-20C fr BiLSTM-CRF ♦
Ortiz S. et al. [137] news newspapers ( hipe) 19C-20C fr, de BiLSTM-CRF ♦
Kristanti et al. [105] news newspapers ( hipe) 19C-20C en, fr BiLSTM-CRF ♦
Provatorova et al. [152] news newspapers ( hipe) 19C-20C de, en, fr BiLSTM-CRF ♦
Todorov et al. [191] news newspapers ( hipe) 19C-20C de, en, fr BiLSTM-CRF ♦
Schweter et al. [173] news newspapers ( hipe) 19C-20C de BiLSTM-CRF ♦
Labusch et al. [107] news newspapers ( hipe) 19C-20C de, en, fr BERT ♦
Ghannay et al. [70] news newspapers ( hipe) 19C-20C fr ♦
Boros et al. [25] news newspapers ( hipe) 19C-20C de, en, fr BERT ♦
Swaileh et al. [184] economy financial yearbooks 20C de, fr BiLSTM-CRF
Yu et al. [203] history state official books 1 bce-17C zh BERT
Hubková et al. [91] news newspapers 19C-20C cz BiLSTM
Table 5. Historical NER literature overview. Papers are grouped by family of approaches and ordered by
publication year. ‘ Comp. ’ stands for comparable and denotes works whose results are obtained on same test
sets.26 Ehrmann et al.
English data, and a custom one trained on 600 articles of the Trove collection (the time period of the
sample is not specified). Interestingly, the model trained on in-domain data did not outperform the
default one, and both yielded F-scores around 75%forPerson andLocation , with a drop below 50%for
Organisation . Neudecker et al. [ 134] focused on newspaper material in French, German and Dutch
from the Europeana collection [ 132], on which they trained a Stanford CRF model with additional
gazetteers. The 4-fold cross-evaluation yielded F-scores in the range of 70-80% for Dutch and French,
while no results were reported for German. For both languages, recall was significantly lower than
precision. Working on Finnish historical newspapers, Ruokolainen et al. [ 169] considered Person
andLocation and trained the Stanford CRF classifier on manually corrected OCRed material, with
large gazetteers covering inflected forms. The model gave satisfying performances with F-scores of
87%(location) and 80%(person) on a test set taken from the same manually corrected data, and of
78%and 71%on non-corrected OCR texts (with recall being lower than precision). This time on
French, and taking advantage of the Quaero Old Press corpus, Galibert et al. [ 67] organised a small
evaluation campaign where three anonymous systems participated. Stochastic systems performed
best (especially on noisy entities), with an F-score of 65 .2%across all types (person, location and
organisation). Also working on French newspapers in the context of HIPE-2020, Elvaigh et al. [ 58]
(slightly) fine-tuned the CRF baseline provided by the organisers and reached 66%on all types
(exact match), two points more than the baseline.
Going back in time, Aguilar et al. [ 1] experimented NER on manually transcribed Latin medieval
charters from the 10C to 13C. Focusing on person and place names, they used dedicated pre-
processing and trained a CRF classifier using the Wapiti toolkit.36Results are remarkable, on
average in the 90%for both types, certainly due to the regularity of the documents in terms of
names, naming conventions, context and overall structure.
Finally, Passaro et al. [ 141] attempted to extract entities from WWI and WWII Italian official
war bulletins. They focused on the traditional entity types, plus Military organisations ,Ships and
Airplanes . The Stanford system was trained (without gazetteers) on semi-automatically annotated
data from the two periods as well as on contemporary Italian news, and various experiments mixing
in- vs. out-of-time data were carried out. Results showed that performances are highest when
the model is trained on data close in time, that entities of type Location are systematically better
recognised, and that custom types (ships, military organisations, etc.) are poorly recognised.
Conclusion on traditional machine learning approaches. Overall, the availability of machine
learning-based NER systems that could either be applied as such or trained on new material greatly
fostered a second wave of experiments on historical documents. Settings are quite diverse, and
so are the performances, but F-scores are usually in the order of 60−70%, which is significantly
lower than those usually obtained on contemporary material (frequently in the 90%). The Stanford
CRF classifier is by far the most commonly used, as well as CRF in general. Not surprisingly,
performances are higher when systems are trained on in-domain material.
6.3 Deep Learning Approaches
Latest developments in historical NER are dominated by deep learning techniques which have
recently shown state-of-the-art results for modern NER. Deep learning-based sequence labelling
approaches rely on word and character distributed representations and learn sentence or sequence
features during end-to-end training. Most models are based on BiLSTM architectures or self-
attention networks, and use a CRF layer as tag decoder to capture dependencies between labels (see
Section 3.4.3). Building on these results, much work attempt to apply and/or adapt deep learning
approaches to historical documents, under different settings and following different strategies.
36https://wapiti.limsi.fr/Named Entity Recognition and Classification on Historical Documents: A Survey 27
6.3.1 Preliminary comments. Let us begin with some observations on the main lines of research.
In a feature learning context the crucial point is, by definition, the capacity of the model to learn
or reuse appropriate knowledge for the task at hand. Given a situation of time and domain shifts
and of resource scarcity, what is at stake for neural-based historical NER approaches is to capture
historical language idiosyncrasies (including OCR noise) and to adequately leverage previously
learned knowledge — a process made increasingly possible with the usage of pre-trained language
models in a transfer learning context. Transfer learning (TL) refers to a set of methods which
aims at leveraging knowledge from a source setting and adapting it to a target setting [ 140]. TL is
not new in NLP but was recently given considerable momentum, in particular sequential transfer
learning where the source task (e.g. language modeling) differs from the target task (e.g. NER).
In this supervised TL setting, a widely used process is to first learn representations on a large
unlabelled corpus (source), before adapting them to a specific task using labelled data (target). The
previously learned model can be adapted to the target task in different ways, the most frequent
being weight adaptation, where pre-trained weights are either kept unchanged (‘frozen’) and used
as features in the downstream model (feature extraction), or fine-tuned to the target task and used
as initialisation in the downstream model (fine-tuning) [168].
To date, most DL approaches to historical NER have primarily focused on experimenting with
a) different input representations, that is to say embeddings of different granularity (character,
sub-word, word), learned at the type or token level (static vs. contextualised) and derived from
domain data or not (in vs. out-of-domain), and b) different transfer learning strategies. Those aspects
are often intermingled in various experiments reported in the literature, which does not easily lend
itself to a clear-cut narrative outline. The discussion which follows is organised according to the
demarcation line ‘words vs. words-in-context’, complemented with observations on TL settings and
types of networks. However imperfect this line is, it reflects the recent evolution of incorporating
more context and of testing all-round language models in historical settings. As a complement,
and in order to frame further the discussion, we identified a set of key research questions from the
types of experiments reported in publications, summarised in Table 6.
6.3.2 Approaches based on static embeddings. First attempts are based on state-of-the-art BiLSTM-
CRF and investigate the transferability of various types of pre-trained static embeddings to historical
material. They all use traditional CRFs as baseline.
Focusing on location names in 19-20C English travelogues,37Sprugnoli [ 178] compares two
classifiers, Stanford CRF and BiLSTM-CRF, and experiment with different word embeddings: GloVe
embeddings, based on linear bag-of-words contexts and trained on Common Crawl data [ 143],
Levy and Goldberg embeddings, produced from the English Wikipedia with a dependency-based
approach [ 115], and fastText embeddings, also trained on the English Wikipedia but using sub-word
information [ 21]. Additionally to these pre-trained vectors, Sprugnoli trains each embedding type
afresh on historical data (a subset of the Corpus of Historical American English), ending up with
3×2 input options for the neural model. Both classifiers are trained on a relatively small labelled
corpus. Results show that the neural approach performs systematically and remarkably better than
CRF, with a difference ranging from 11 to 14 F-score percentage points, depending on the word
vectors used (best F-score is 87.4 %). If in-domain supervised training improves the F-score of the
Stanford CRF module, it is worth noting that the gain is mainly due to recall, the precision of the
English default model remaining higher. In this regard, the neural approach shows a better P/R
balance across all settings. With respect to embeddings, linear bag-of-words contexts (GloVe) prove
to be more appropriate (at least in this context), with its historical embeddings yielding the highest
scores across all metrics (fastText following immediately after). A detailed examination of results
37Corpus presented in Section 5.2.2.28 Ehrmann et al.
Research questions Experiments Publication
Input representation
Which type of embedding is best?
Test different static embedding algorithms [178]
Test different static embedding granularity [162]
Use modern static embeddings (word2vec, fastText) [91]
Use modern char-level LM embeddings (Flair) [184]
Use modern word-level LM embeddings (BERT, ELMo) [70, 152, 203]
Uses stack of modern embeddings [105, 137, 162]
Transfer learning
How well modern embeddings can transfer to historical texts?
What is the impact of in-domain embeddings?
Is more task-specific labelled data more helpful than big or in-domain LMs?
Test modern vs. historical static embeddings [158]
Test modern vs. historical char-level LM embeddings [41, 101, 137, 172, 173]
Test modern vs. historical word-level LM embeddings [2, 108, 172]
Test stack of embeddings [2, 25, 108, 172, 173, 191]
Test feature extraction (frozen) vs. fine-tuning [91, 152, 162]
Test different training corpus sizes [2, 105, 158]
Test cross-corpus model application [25, 105, 108, 158, 191]
Test cross-corpus model training [158]
Neural architecture
How neural approaches compare to traditional CRFs?
What is the best neural architecture with which decoder?
Compare BiLSTM and traditional CRF [137, 158, 162, 178]
Compare CRF decoder vs. softmax decoder [162]
Compare BiLSTM and LSTM [91]
Test single vs. multitask learning [162, 191]
Compare transformers and BiLSTM [25]
Table 6. Synthetic view of DL experiments mapped with research questions.
reveals an uneven impact of in-domain embeddings, leading either to higher precision but lower
recall (Levy and GloVe), or higher recall but lower precision (fastText and GloVe). Overall, this
work shows the positive impact of in-domain training data: the BiLSTM-CRF approach, combined
with in-domain training set and in-domain historical embeddings, systematically outperforms the
linear CRF classifier.
In the context of reference mining in the arts and humanities, Rodriguez et al. [ 162] also inves-
tigate the benefit of BiLSTM over traditional CRFs, and of multiple input representations. Their
experiments focus on three architectural components: input layer (word and character-level word
embeddings), prediction layers (Softmax and CRF), and learning setting (multi-task and single-task).
Authors consider a domain-specific tagset of 27 entity types covering reference components (e.g.
author, title, archive, publisher) and work with 19-21C scholarly books and journals featuring a
wide variety of referencing styles and sources.38While character-level word embeddings, likely to
38Corpus presented in Section 5.2.2Named Entity Recognition and Classification on Historical Documents: A Survey 29
help with OCR noise and rare words, are learned either via CNN or BiLSTM, word embeddings
are based on word2vec and are tested under various settings: present or not, pre-trained on the in-
domain raw corpus or randomly initialised, and frozen or fined-tuned on the labelled corpus during
training. Among those settings, the one including in-domain word embeddings further fine-tuned
during training and CRF prediction layer yields the best results ( 89 .7%F-score). Character-level
embeddings provide a minor yet positive contribution, and are better learned via BiLSTM than
with CNN. The BiLSTM architecture outperforms the CRF baseline by a large margin (+ 7%), except
for very infrequent tags. Overall, this work confirms the importance of word information (rather
in-domain, though here results with generic embeddings were not reported) and the remarkable
capacities of a BiLSTM network to learn features, better decoded by a CRF classifier than a softmax
function.
Working with Czech historical newspapers,39Hubková et al. [ 91] target the recognition of
five generic entity types. Authors experiment with two neural architectures, LSTM and BiLSTM,
followed by a softmax layer. Both are trained on a relatively small labelled corpus (4k entities) and
fed with modern fastText embeddings (as released by the fastText library) under three scenarios:
randomly initialised, frozen, and fine-tuned. Character-level word embeddings are not used. Results
show that the BiLSTM model based on pre-trained embeddings with no further fine-tuning performs
best ( 73%F-score). Authors do not comment on the performance degradation resulting from fine-
tuning, but one reason might be the small size of the training data.
Rather than aiming at calibrating a system to a specific historical setting, Riedl et al. [ 158]
adopt a more generic stance and investigate the possibility of building a German NER system
that performs at the state of the art for both contemporary and historical texts. The underlying
question—whether one type of model can be optimised to perform well across settings— naturally
resonates with the needs of cultural heritage institution practitioners (see also Schweter et al. [ 172]
and Labush et al. [ 108] hereafter). Experimental settings consist of: two sets of German labelled
corpora, with large contemporary datasets (CoNNL-03 and GermEval) and small historical ones
(from the Friedrich Temann and Austrian National library); two types of classifiers, CRFs (Stanford
and GermaNER) and BiLSTM-CRF; finally, for the neural system, usage of fastText embeddings
derived from generic (Wikipedia) and in-domain (Europeana corpus) data. On this base, authors
perform three experiments. The first investigates the performances of the two types of systems on
the contemporary datasets. On both GermEval and CoNNL, the BiLSTM-CRF models outperform the
traditional CRF ones, with Wikipedia-based embeddings yielding better results than the Europeana-
based ones. It is noteworthy that the GermaNER CRF model performs better than the LSTM of
Lample et al. [ 110] on CoNLL-03, but suffers from low recall compared to BiLSTM. The second
experiment focuses on all-corpora crossing, with each system being trained and evaluated on all
possible combinations of contemporary and historical corpora pairs. With no surprise, best results
are obtained when models are trained and evaluated on the same material. Interestingly, CRFs
perform better than BiLSTM in the historical setting (i.e. train and test sets from historical corpora)
by quite a margin, suggesting that although not optimised for historical texts, CRFs are more robust
than BiLSTM when faced with small training datasets. The type of embeddings (Wikipedia vs.
Europeana) plays a minor role in the BiLSTM performance in the historical setting. Ultimately, the
third experiment explores how to overcome this neural net dependence on large data with domain
adaptation transfer learning: the model is trained on a contemporary corpus until convergence and
then further trained on a historical one for a few more epochs. Results show consistent benefits
for BiLSTM on historical datasets (ca. +4 F-score percentage points). In general, main difficulties
relate to OCR mistakes and wrongly hyphenated words due to line breaks, and to the Organisation
39Corpus presented in Section 5.2.130 Ehrmann et al.
type. Overall, this work shows that BiLSTM and CRF achieve similar performances in a small-data
historical setting, but that BiLSTM-CRF outperforms CRF when supplied with enough data or in a
transfer learning setting.
This first set of work confirms the suitability of the state-of-the-art BiLSTM-CRF approach for
historical documents, with the major advantage of not requiring feature engineering. Provided
that there is enough in-domain training data, this architecture obtains better performances than
traditional CRFs (the latter performing on par or better otherwise). In-domain pre-training of static
word embeddings seems to contribute positively, although to various degrees depending on the
experimental settings and embedding types. Sub-word information (either character embeddings
or character-based word embeddings) also appears to have positive effect.
6.3.3 Approaches based on character-level LM embeddings. Approaches described above rely on
static, token-level word representations which fail to capture context information. This drawback
can be overcome by context-dependent representations derived from the task of modelling language,
either as distribution over characters, such as the Flair contextual string embeddings [ 3], or over
words, such as BERT [ 43] and ELMo [ 144] (see Section 3.3.3). Such representations have boosted
performances of modern NER and are also used in the context of historical texts. This section
considers work based on character-based contextualised embeddings (flair).
In the context of the CLEF-HIPE-2020 shared task [ 53], Dekhili et al. [ 41] proposed different
variations of a BiLSTM-CRF network, with and without the in-domain HIPE flair embeddings
and/or an attention layer. The gains of adding one or the other or both are not easy to interpret,
with uneven performances of the model variants across NE types. Their overall F-scores range from
62%to65%under the strict evaluation regime. For some entity types the CRF baseline is better than
the neural models, and the benefit of in-domain embeddings is overall more evident than the one
of the attention layer (which proved more useful in handling metonymic entities).
Kew et al . [101] address the recognition of toponyms in an alpine heritage corpus consisting of
over 150 years of mountaineering articles in five languages (mainly from the Swiss and British
Alpine Clubs). Focusing on fine-grained entity types (city, mountain, glacier, valley, lake, and cabin),
the authors compare three approaches. The first is a traditional gazetteer-based approach completed
with a few heuristics which achieves high precision across types ( 88%P,73%F-score), and even very
high precision ( >95%) for infrequent categories with regular patterns. Suitable for reliable location-
based search but suffering from low recall, this approach is then compared with a BiLSTM-CRF
architecture. The neural system is fed with stacked embeddings composed of in-domain contextual
string embeddings pre-trained on the alpine corpus concatenated with general-purpose fastText
word embeddings pre-trained on web data, and trained on a silver training set containing 28k
annotations obtained via the application of the gazetteer-based approach. The model leads to
an increase of recall for the most frequent categories, without degrading precision scores ( 76%
F-score). This shows the generalisation capacity of the neural approach in combination with context-
sensitive string embeddings and given sufficient training data. Finally, authors experiment with
crowd-corrected annotations and observe that already a small number of corrections on the silver
data has a positive impact (+3 F-score percentage point).
Swaileh et al . [184] target even more specific entity types in French and German financial
yearbooks from the first half of 20C. They apply a BiLSTM-CRF network trained on custom data
and fed with modern flair embeddings. Results are very good (between 85%to95%F-score depending
on the book sections), with the CRF baseline and the BiLSTM model performing on par for French
books, and BiLSTM being better than CRF for the German one, which has a lower OCR quality.
Overall, these performances can be explained by the regularity of the structure and language as
well as the quality of the considered material, resulting in stable contexts and non-noisy entities.Named Entity Recognition and Classification on Historical Documents: A Survey 31
6.3.4 Approaches based on word-level LM embeddings. The release of pre-trained contextualised
language model-based word embeddings such as BERT (based on transformers) and ELMo (based on
LSTM) pushed further the upper bound of modern NER performances. They show promising results
either in replacement or in combination with other embedding types, and offer the possibility of
being further fine-tuned [ 116]. If they are becoming a new paradigm of modern NER, the same
seems to be true for historical NER.
Using pre-trained modern embeddings. We first consider work based on pre-trained modern
LM-based word embeddings (BERT or ELMo) without extensive comparison experiments. They
make use of BiLSTM or transformer architectures.
Working on the “Chinese Twenty-Four Histories”, a set of Chinese official history books covering
a period from 3000 BCE to 17C, Yu et al. [ 203] face the problems of the complexity of classical
Chinese and of the absence of appropriate training data in their attempt to recognise Person and
Location . Their BiLSTM-CRF model is trained on a NE-annotated modern Chinese corpus and
makes use of modern Chinese BERT embeddings in a feature extraction setting (frozen). Evaluated
on a (small) dataset representative of the time span of the target corpus, the model achieves
relatively good performances (from 72%to82%F-score depending on the book), with a pretty good
P/R balance, better results for Location than for Person , and on recent books. Given the complete
‘modern’ setting of embeddings and training labelled data, those results shows the benefit of large
LM-based embeddings—keeping in mind the small size of the test set and perhaps the regularity of
entity occurrences in the material, not detailed in the paper.
Also based on the bare usage of state-of-the-art LM-based representations is a set of work from
the HIPE-2020 evaluation campaign. These work tackle the recognition of five entity types in about
200 years of historical newspapers in French, English, and German.40The task included various NER
settings, however only the coarse literal NE recognition is considered here. Ortiz Suárez et al . [137]
focused on French and German. They first pre-process the newspaper line-based format (or column
segments) into sentence-split segments before training a BiLSTM-CRF model using a combination
of modern static fastText and contextualised ELMo embeddings as input representations. They
favoured ELMo over BERT because of its capacity to handle long sequences and its dynamic
vocabulary thanks to its CNN character embedding layer. In-domain fastText embeddings provided
by the organisers were tested but performed lower. Their models ranked third on both languages
during the shared task, with strict F-score of 79%and 65%for French and German respectively.
The considerably lower performance of their improved CRF baseline illustrates the advantage of
contextual embeddings-based neural models. Ablation experiments on sentence splitting showed
an improvement of 3.5 F-score percentage points on French data (except for Location ) confirming
the importance of proper context for NER neural tagging.
Running for French and English, Kristanti et al. [ 105] also make use of a BiLSTM-CRF relying
on modern fastText and ELMo emddings. In the absence of training set for English, authors use
the CoNLL-2012 corpus, while for French the training data is further augmented with another NE-
annotated journalistic corpus from 1990, which proved to have positive impact. They scored at 63%
and 52%in terms of strict F-score for French and English respectively. Compared to the French results
of Ortiz Suàez et al., Kristanti et al. use the same French embeddings but a different implementation
framework and different hyper-parameters, and does not apply sentence segmentation.
Finally, still within the HIPE-2020 context, two teams tested pre-trained LM embeddings with
transformer-based architectures. Provatorova et al . [152] proposed an approach based on the fine-
tuning of BERT models using Huggingface’s transformer framework for the three shared task’s
languages, using the cased multilingual BERT base model for French and German and the cased
40Corpus presented in Section 5.2.1.32 Ehrmann et al.
monolingual BERT base model for English. They used the CoNLL-03 data for training their English
model, the HIPE data for the others, and additionally set up a majority vote ensemble of 5 fine-tuned
model instances per language in order to improve the robustness of the approach. Their models
achieved F-scores of 68%,52% and 47% for French, German and English respectively. Ghannay
et al. [70] used CamemBERT, a multi-layer bidirectional transformer similar to ROBERTa [ 119,124]
initialised with a pre-trained modern French CamemBERT and completed with a CRF tag decoder.
This model obtained the second-best results for French with 81%strict F-score.
Even when learned from modern data, pre-trained LM-based word embeddings encode rich prior
knowledge that effectively support neural models trained on (usually) small historical training sets.
As for HIPE-related systems, it should be noted that word-level LM embeddings systematically
lead to slightly higher recall than precision, demonstrating their powerful generalisation capacities,
even on noisy texts.
Using modern and historical pre-trained embeddings. As for static embeddings, it is logical to expect
higher performances from LM-embeddings when pre-trained on historical data, in combination
with modern ones or not. The set of work reviewed here explores this perspective.
Ahmed et al . [2] work on the recognition of universal and domain-specific entities in German
historical biodiversity literature.41They experiment with two BiLSTM-CRF implementations (their
own and Flair framework) which both use modern token-level German word embeddings and
are trained on the BIOfid corpus. Experiments consist in adding richer representations (modern
Flair embeddings, additionally completed by newly trained ELMo embeddings or BERT base
multilingual cased embeddings) or adding more task-specific training data (GermEval, CoNLL-03
and BIOfid). Models perform more or less equally, and authors explained the low gain of in-domain
ELMo embdedings by the small size of the training data (100k sentences). Higher gains come with
larger labelled data, however the absence of ablation tests hinders the complete understanding
of the contribution of the historical part of this labelled data, and the use of two implementation
frameworks does not warrant full results comparability.
Both Schweter et al. [ 172] and Labusch et al. [ 108] build on the work of Riedl et al. [ 158] and
try to improve NER performances on the same historical German evaluation datasets, thereby
constituting (with HIPE-2020) one of the few sets of comparable experiments. Schweter et al. seek
to offset the lack of training data by using only unlabelled data via pre-trained embeddings and
language models. They use the Flair framework to train and combine (“stack”) their language
models, and to train a BiLSTM-CRF model. Their first experiment consists in testing various static
word representations, with: character embeddings learned during training, fastText embeddings
pre-trained on Wikipedia or Common Crawl (with no sub-word information), and the combination
of all of these. While Riedl et al. experimented with similar settings (character embeddings and
pre-trained modern and historical fastText embeddings), it appears that combining Wikipedia and
Common Crawl embeddings leads to better performances, even higher than the transfer learning
setting of Riedl et al. using more labelled data. As a second experiment, Schweter et al. use pre-
trained LM embeddings: flair embeddings newly trained on two historical corpora having temporal
overlaps with the test data, and two modern pre-trained BERT models (multilingual and German).
On both historical test sets, in-domain LMs yield the best results (outperforming those of Riedl et
al.), all the more so when the temporal overlap between embedding and task-specific training data
is large. This demonstrates that the selection of the language model corpus plays an important role,
and that unlabelled data close in time might have more impact than more (and difficult to obtain)
labelled data.
41Corpus presented in Section 5.2.2Named Entity Recognition and Classification on Historical Documents: A Survey 33
With the objective of developing a versatile approach that performs decently on texts of different
epochs without intense adaptation, Labusch et al. [ 108] experiment with BERT under different
pre-training and fine-tuning settings. In a nutshell, they apply a model based on multilingual
BERT embeddings, which is further pre-trained on large OCRed historical German unlabelled
data (the Digital Collection of the Berlin State Library) and subsequently fine-tuned on several
NE-labelled datasets (CoNLL-03, GermEval, and the German part of Europeana NER corpora).
Tested across different contemporary/historical dataset pairs (similar to the all-corpora crossing of
Riedl et al. [ 158]), it appears that additional in-domain pre-training is most of the time beneficial
for historical pairs, while performances worsen on contemporary ones. The combination of several
task-specific training datasets has positive yet less important impact than BERT pre-training, as
already observed by Schweter et al. [ 172]. Overall, this work shows that an appropriately pre-trained
BERT model delivers decent recognition performances in a variety of settings. In order to further
improve them, authors purpose to use the BERT large instead of the BERT base model, to build
more historical labelled training data, and to improve the OCR quality of the collections.
The same spirit of combinatorial optimization drove the work of Todorov et al. [ 191] and
Schweter et al. [ 173] in the context of HIPE-2020. Todorov et al. build on the bidirectional LSTM-
CRF architecture of Lample et al. and introduce a multi-task approach by splitting the top layers
for each entity type. Their general embedding layer combines a multitude of embeddings, on
the level of characters, sub-words and words; some newly trained by the authors, as well as pre-
trained BERT and HIPE’s in-domain fastText embeddings. They also vary the segmentation of
the input: line segmentation, document segmentation as well as sub-document segmentation for
long documents. No additional NER training material was used for German and French, while for
English, the Groningen Meaning Bank42was adapted for training. Results suggest that splitting
the top layers for each entity type is not beneficial. However, the addition of various embeddings
improves the performance, as shown in the very detailed ablation test report. In this regard,
character-level and BERT embeddings are particularly important, while in-domain embeddings
contribute mainly to recall. Fine-tuning pre-trained embeddings did not prove beneficial. Using
(sub-)document segmentation clearly improved results when compared to the line segmentation
found in newspapers, emphasising once again the importance of context. Post-campaign F-scores
for coarse literal NER are 75%and 66%for French and German (strict setting). English experiments
yielded poor results, certainly due to the time and linguistic gaps between training and test data,
and the pretty bad OCR quality of the material (in the same way as for Provatorova et al . [152] and
Kristanti et al. [105]).
For their part, Schweter et al. [ 173] focused on German and experimented with ensembling
different word and subword embeddings (modern fastText and historical self-trained and HIPE
flair embeddings), as well as transformer-based language models (trained on modern and historical
data), all integrated by the neural Flair NER tagging framework [ 3]. They used a state-of-the-
art BiLSTM with an on-top CRF layer as proposed by [ 89], and perform sentence splitting and
hyphen normalisation as pre-processing. To identify the optimal combination of embeddings
and LMs, authors first selected the best embeddings for each type before combining them. Using
richer representations (fastText<flair<BERT) leads to better results each time. Among the options,
Wikipedia fastText embeddings proved better than the Common Crawl ones, suggesting that similar
data (news) is more beneficial than larger data for static representations; HIPE flair embeddings
proved better than other historical ones, likely because of their larger training data size and data
proximity; and BERT LM trained on large data proved better than the one trained on historical
42https://gmb.let.rug.nl/34 Ehrmann et al.
(smaller) data. The best final combination includes fastText and BERT, leading to 65%F-score on
coarse literal NER (strict setting).
Finally, Boros et al . [25] also tackled NER tagging for HIPE-2020 in all languages and achieved
best results. They used a hierarchical transformer-based model [ 196] built upon BERT in a multi-
task learning setting. On top of the pre-trained BERT blocks (multilingual BERT for all languages,
additionally Europeana BERT for German43and CamemBERT for French [ 124]), two task-specific
transformer layers were optionally added to alleviate data sparsity issues, for instance out-of-
vocabulary words, spelling variations, or OCR errors in the HIPE dataset. A state-of-the-art CRF
layer was added on top in order to model the context dependencies between entity tags. For base
BERT with a limited context of 512 sub-tokens, documents are too long and newspaper lines are
too short for proper contextualization. Therefore, an important pre-processing step consisted in the
reconstruction of hyphenated words and in sentence segmentation. For the two languages with in-
domain training data (French and German), their best run consisted in BERT fine-tuning, completed
with the two stacked transformer blocks and the CRF layer. For English without in-domain training
data, two options for fine-tuning were tested: a) training on monolingual CoNLL-03 data, and b)
transfer learning by training on the French and German HIPE data. Both options worked better
without transformer layers, and training on the French and German HIPE data led to better results.
Final F-scores for coarse literal NER were 84%,79% and 63% for French, German and English
respectively (strict setting).
Conclusion on deep learning approaches. What conclusions can be drawn from all this? First,
the twenty or so papers reviewed above illustrate the growing interest of researchers and practi-
tioners from different fields in the application of deep learning approaches to NER on historical
collections. Second, it is obvious that these many publications also equate with a great diversity
in terms of document, system and task settings. Apart from the historical German [ 108,158,172]
and HIPE papers, most publications use different datasets and evaluation settings, which pre-
vents result comparison; what is more, the sensitivity of DL approaches to experimental settings
(pre-processing, embeddings, hyper-parameters, hardware) usually undermines any attempt to
compare or reproduce experiments, and often leads to misconceptions about what works and what
does not [ 202]. As shown in the DL literature review above, what is reported can sometimes be
contradictory. However, and with this in mind, a few conclusions can be drawn:
–State-of-the-art BiLSTM architectures achieve very good performances and largely outper-
form traditional CRFs, except in small data contexts and on very regular entities. As inference
layer, CRF is a better choice than softmax (also confirmed by Yang et al . [202] ). Yet, in the
fast-changing DL landscape, transformer-based networks are already taking over BiLSTM.
–Character and sub-word information is beneficial and helps the model to deal with OOV
words, presumably historical spelling variations and OCR errors. CNN appears to be a better
option than LSTM to learn character embeddings.
–As for word representation, the richer the better. The same neural architecture performs
better with character or word-based contextualised embeddings than with static ones, and
even better with stacked embeddings. The combination of flair or fastText embeddings plus
a BERT language model seems to provide an appropriate mix of morphological and lexical
information. Contextualised representations also have positive impact in low resource setting.
–Pre-trained modern embeddings prove to transfer reasonably well to historical texts, even
more when learned on very large textual data. As expected, in-domain embeddings contribute
positively to performances most of the time, and the temporal proximity between the corpora
43https://github.com/stefan-it/europeana-bertNamed Entity Recognition and Classification on Historical Documents: A Survey 35
from which embeddings are derived and the targeted historical material seems to play an
important role. Although a combination of generic and historical prior knowledge is likely to
increase performances, what is best between very large modern vs. in-domain LMs remains
an open question.
–Careful pre-processing of input text (word de-hyphenation, sentence segmentation) in order
to work with valid linguistic units appears to be a key factor.
Ultimately, apart from clearly outperforming traditional ML and rule-based systems, the most
compelling aspect of DL approaches is certainly their transferability; if much still need to be
investigated, the possibility of having systems performing (relatively) well across historical settings—
or a subset thereof—seems to be an achievable goal.
7 STRATEGIES TO DEAL WITH SPECIFIC CHALLENGES
We report here on the main strategies implemented by the different types of approaches to overcome
OCR noise, adapt to language shifts and deal with lack of resources. Table 7 provides a synthetic
view of the challenges, their impact, and the possible solutions to address them.
7.1 Dealing with noisy input
There exist two main strategies to deal with OCR and OLR noise: adapting the input, i.e. correcting
the text before parsing it, or adapting the tool, i.e. making the NER system capable of dealing with
the noise. Let us recall here that OCR and OLR noise mostly correspond to: misrecognised characters,
erroneously truncated or connected word parts, spurious word hyphenations and incorrect mix of
textual segments, all of these translating into OOV words and/or inconsistencies affecting both
entities and their context.
The first strategy corresponds to OCR/OLR post-correction and aims at recovering correct word
forms and rebuilding linguistically motivated token sequences. Such processes depend on the
specifics of each input (e.g. in terms of layout, typographic conventions, presence of marginalia)
and are not easy to implement given the countless erroneous punctuation marks added by OCR
and the subtle difference between soft and hard hyphens, an information often lost through the
different digital versions of a document. In this context, most work apply a mix of well-known
and ad hoc correction strategies based on corpus examination, including: a) correction of the ‘long
s’, an archaic from of the ‘s’ letter systematically confused with ‘f’ by OCR engines [ 6]; b) word
de-hyphenation, which consists in removing the end-of-line soft hyphens and checking the validity
of the resulting word form [ 6,44]; c) word OCR post-correction based on the edit distance between
input tokens and a list of most frequent OOV words manually corrected (allowing to correct
the majority of mistakes) [ 44]; d) application of a generic spelling correction system [ 93]; and e)
application of sentence segmentation [ 25,137]. Word de-hyphenation and sentence segmentation
(or an approximation of it, depending on the quality of the input) are beneficial for all types of
systems but are particularly critical for neural-based systems. Word correction have positive albeit
moderate and irregular impact on the performances [ 44], which illustrates the difficulty of OCR
post-correction and raises the question of under which conditions it is most beneficial. Using a
generic spelling correction system, Huynh et al. [ 93] precisely leave aside the fine-tuning of OCR
correction to focus, instead, on the question of when to apply it. They show that post-OCR correction
is most beneficial with character and word-level error rates above 3% and 20% respectively, while
it degrades NER performances for lower rates, due to spurious corrections. Overall, input noise
correction or reduction is beneficial but should be adjusted according to the type and importance
of noise.36 Ehrmann et al.
Challenges Impact Possible solutions
Noisy input
sparser feature space,
low recall.OCR post-correction
string similarity
historical LMs
in-domain training data
sub-word tokenisation•OOV
•broken token sequences
Dynamics of language
sparser feature space,
low recall.normalisation
historical LMs
in-domain training data•spelling variations
•name irregularities
•entity drift
Lack of resources
limited learning capacities,
limited system comparison.transfer learning
active learning
data augmentation
resource sharing•inappropriate typologies
•lack of NE-annotated corpora
•paucity of historical LMs
Table 7. NER on historical documents: main challenges, their impact, and possible solutions.
Another approach is to leave the input untouched but to provide the system with relevant
information about the noise, mainly in the form of embedded language representations for neural-
based approaches. In this regard, word embeddings computed at the character and sub-words
levels (fastText [ 21]) as well as character-level LM embeddings (flair [ 4]) are particularly efficient
in dealing with miss-recognised characters. For example, a BiLSTM-CRF model based on flair
historical embeddings could correctly recognise and label T«i*louse (Toulouse) and Caa.Qrs (Ca-
hors) as Location , and o˚an (Jean) asPerson [20]. Beside input representations, transformer-based
architectures such as BERT integrate sub-word tokenizers based on e.g. Byte-Pair Encoding [ 174]
or WordPiece [ 199] algorithms. Such tokenisation allows to consider word pieces not present in
the LM vocabulary, thereby learning representations of OOV words. This is however not the total
answer since resulting sub-words depend on the anatomy of the misspelling: if character insertion
mostly results in known sub-word units, substitution or deletion produce uncommon units which
still cause difficulties [ 183]. Boros et al. [ 24] (outperforming their previous work [ 25]) carried out an
in-depth error analysis of the performances of different variants of a BERT-based system augmented
(or not) with extra transformer layers, considering among others the percentage of OOV words and
the severeness of OCR noise in entity mentions. Conclusions are that the representation capacity
of the extra layers is beneficial for both recall and precision: while misspelled entities are better
recognised in general, the system with extra layers does not over-predict entities, as is the case for
the non-augmented system. Authors also highlight the possible over-fitting of the base model on
OCR-related patterns in frequent entities, a question on which further research is necessary.
In general, language models and transformers thus appear as good options to deal with noisy
input with the least effort, but the effectiveness of targeted correction heuristics should not be
underrated. Let us conclude with Fig. 2, which illustrates the results of three different systems on
noisy texts.
7.2 Dealing with dynamics of language
Similarly to noise, adaptation to dynamics of language aims at reducing the gap between the text to
analyse and the knowledge of the system. Let us recall the issues at stake here: historical spelling
variations, evolving naming conventions, and entity drift. There is no single and clear-cut answer
to each of these issues, but rather a set of possibilities which can help address them.Named Entity Recognition and Classification on Historical Documents: A Survey 37
NE GT -CRF -BiLSTM -BERT -PERSON
English: [who] was the general Chapedelaine, who was never mentioned in [the military annals]PERSONPERSONétaitle généraiChapecUlaine, dontiln'ajamaisétéquestion dansétaitle généraiChapecUlaine, dontiln'ajamaisétéquestion dansétaitle généraiChapecUlaine, dontiln'ajamaisétéquestion dansétaitle généraiChapecUlaine, dontiln'ajamaisétéquestion dans(1)
NE GT -CRF -BiLSTM -BERT -PRODUCT
PRODUCTEnglish: "[the] attacks that the Wäschfra [has taken the liberty to conduct] against [...]". Angriffe, welches sichDie „3Sd)cb.fra" gegenAngriffe, welches sichDie „3Sd)cb.fra" gegenAngriffe, welches sichDie „3Sd)cb.fra" gegenAngriffe, welches sichDie „3Sd)cb.fra" gegen(2)
Fig. 2. Results of a CRF, BiLSTM-CRF and BERT-based NER systems on excerpts from the French Swiss
Gazette de Lausanne of August 4 1818, p.4 (1), and from the German Luxembourgian Luxemburger Wort of
July 21 1868, p.2 (2) (HIPE data), compared to the ground truth (NE GT).
As previously mentioned, rule-based and traditional ML systems using gazetteers often suffer
from low recall due to vocabulary mismatches between name variants and gazetteer entries. Several
approaches can help alleviate this issue. First, the lookup conditions can be loosened using a string
similarity measure with an empirically determined threshold. In this respect, Borin et al. [ 23]
showed that allowing a small Levenshtein distance between input tokens and gazetteers entries
allows to capture orthographic variations between 19C and 20C Swedish names and thus to increase
the performances of their rule-based system, particularly recall (+5 percentage points). Another
option consists in normalising historical spellings using transformation rules, either manually
crafted, such as in Neudecker [ 132] and Platas et al. [ 150], or automatically acquired on aligned
texts such as in Kogkitsidou et al. [ 104]. Results obtained on normalised versions of texts or entities
are usually better, though a somehow contrasted picture emerges depending on the system, the
type of texts, and the time period — as per OCR post-correction.
When trying to adapt the knowledge of the system rather than the input, the key factor is,
not surprisingly, temporal proximity. Here word embeddings and language models derived from
temporally close corpora seem to be better able to capture historical language idiosyncrasies,
including spelling variations [ 108,172]. There is no clear evidence yet regarding what is best
between pre-training from scratch on historical texts or fine-tuning a modern model on historical
texts (in a supervised or unsupervised manner) and further research is necessary on this point.
Beyond technicalities, an important aspect to consider when adapting a system to time is its
application scope, i.e. whether it is intended to perform well on a unique target (one document
type, one time period) or across several.
7.3 Dealing with the lack of resources
As emphasised in Section 4.4, the situation of lack of resources is not unique to historical NER and
corresponds here to inappropriate typologies, and lack of labelled and unlabelled historical corpora.
With respect to typologies, one can only adapt and/or define a typology when existing tag sets
are not appropriate [ 2,150,189]. As easy as it may seem, two comments are in order here: first,
this represents a time-consuming process which requires several expertise (in linguistics, in the
historical domain at hands, and in knowledge representation) and needs to be documented, notably
via annotation guidelines. Specifically, phenomena such as nested entities and metonymy did not38 Ehrmann et al.
received much attention in modern NER but are of high interest for humanists’ (re)search needs.
Second, careful attention should be paid to typologies interoperability, without which resources
are mere silos and need an extra mapping step [161].
Several strategies can be adopted to cope with the lack of training data. The most widely used so
far is transfer learning, as described in Section 3.4.3. Another option is active learning, where a ML
system asks an oracle (or a user) to select the most relevant examples to consider, thereby lowering
the number of data points required to learn a model. This is the approach adopted by Erdmann et
al. [60] to recognise entities in various Latin classical texts, based on an active learning pipeline
able to predict how many and which sentences need to be annotated to achieve a certain degree
of accuracy, and later on released as toolkit to build custom NER models for the humanities [ 61].
Finally, another strategy is data augmentation, where an existing data set is expanded via the
transformation of training instances without changing their label. This approach, which has not yet
been deployed in a historical context, has been successfully tested by Dai et al. [ 39] on biomedical
data, where several data augmentation techniques, in isolation or in combination, led to improved
performance, especially with small training datasets.
8 CONCLUSIONS AND OUTLOOK
We presented an extensive survey of research, published in the last 20 years, on the topic of NER
on historical documents. We introduced the main challenges of historical NER, namely document
type and domain variety, noisy input, dynamics of language, and lack of resources. We inventoried
existing resources available for historical NER (typologies and annotation guidelines, annotated
corpora and language representations), and surveyed the approaches developed to date to tackle
this task, paying special attention to how they adapt to historical settings.
What clearly emerges from this review is, first, that research on historical NER has gained real
momentum over the last decade. The availability of machine-readable historical texts coupled with
the recent advances in deep learning has led to increased attention from researchers and cultural
heritage practitioners for what has now become a cornerstone of semantic indexing of historical
documents. Second, the body of research on historical NER started by following state-of-the-art
approaches in modern NER (with rule-based and then traditional ML approaches), before fully
experimenting with the various possibilities arising from diachronic transfer learning with neural-
based approaches. This last development helped increase the performances of NER systems on
historical material with F-scores going from 60-70% on average for rule-based and traditional ML
systems to, for the best neural systems, 80%. As of today, it is therefore possible to design systems
capable of dealing with historical and noisy inputs, whose performances almost compete with
those obtained on contemporary texts. This success, however, should not conceal the progress still
to be made. In this regard, we outline a set of key priorities for the next generation of historical
NER systems:
(1)Transferability . We emphasise that beyond addressing a specific type of document and/or time
period lies the question of systems’ portability across historical settings. While addressing
system adaptability across both time and domain at once might be overly ambitious for
the time being, having systems performing equally well across one or the other is highly
desirable—especially for cultural heritage institutions—and represents to next great challenge.
In this respect, we encourage to pursue and especially systematise further the transfer learning
experiments undertaken so far.
(2)Robustness . Although a great deal of headway has been made to enable systems to deal
with atypical historical inputs, we highlight that OCR/OLR noise and historical spellings are
still the main sources of errors of NER systems on historical texts. One of the way forwardNamed Entity Recognition and Classification on Historical Documents: A Survey 39
includes a better assessment of which type of noise is detrimental and to which extent in
order to devise more systematic and focused strategies to deal with it.
(3)System comparability . The systematic comparison of the advantages and shortcomings of
approaches to historical NER was made difficult because of the variety of settings to which
they applied (domains, time periods, languages) and of the corpora against which they were
evaluated. We stress the importance of gold standards and of shared tasks on historical
material to enable system comparison and drive progress in historical NER.
(4)Finer-grained historical NER . The (re)search interests of scholars go beyond the recognition
of main entity types and we underline the need to carry out finer-grained NER, taking into
account e.g. nested entities, entity name composition, and entity metonymy.
(5)Resource sharing . All recent advances in deep learning were made possible by the availability
of large-scale textual data. While the sharing of such resources has just begun, we emphasise
the need for access to large-scale historical textual data or to language models derived thereof.
This sharing should also extend to typologies, annotation guidelines, and training material,
with a special attention to interoperability.
NER on historical documents is an exciting field of research with high added-value for both NLP
researchers and digital scholars. While the first can test the robustness of their approaches and gain
new insights with respect to domain, language and time adaptation, the second can benefit from
more accurate semantic indexing and text understanding of historical material. Lastly, we wish to
mention two facets which, even if not directly related to development of historical NER systems,
should be considered while working on this topic: any large-scale endeavour around historical NEs
should acknowledge ethical and legal obligations related to personal data protection, and is most
likely to be useful if humanities scholarship knowledge and needs are taken into account within an
interdisciplinary framework.
ACKNOWLEDGMENTS
The work of Maud Ehrmann and Matteo Romanello was supported by the Swiss National Science
Foundation under the grants number CR-SII5_173719 ( Impresso - Media Monitoring of the Past)
and number PZ00P1_186033 (only for MR). The work of Ahmed Hamdi, Elvys Linhares Pontes
(now employed at the Trading Central Labs company) and Antoine Doucet was supported by the
European Union’s Horizon 2020 research and innovation program under grant 770299 (NewsEye).
REFERENCES
[1]Sergio Torres Aguilar, Xavier Tannier, and Pierre Chastang. 2016. Named Entity Recognition Applied on a Data Base
of Medieval Latin Charters. the Case of Chartae Burgundiae. In 3rd International Workshop on Computational History
(HistoInformatics 2016) . CEUR Workshop Proc., Krakow, Poland, 67–71. https://hal.archives-ouvertes.fr/hal-02407159/
[2]Sajawel Ahmed, Manuel Stoeckel, Christine Driller, Adrian Pachzelt, and Alexander Mehler. 2019. BIOfid Dataset:
Publishing a German Gold Standard for Named Entity Recognition in Historical Biodiversity Literature. In Proc.
of the 23rd Conference on Computational Natural Language Learning (CoNLL) . ACL, Hong Kong, China, 871–880.
https://doi.org/10.18653/v1/K19-1081
[3]Alan Akbik, Tanja Bergmann, Duncan Blythe, Kashif Rasul, Stefan Schweter, and Roland Vollgraf. 2019. FLAIR:
An Easy-to-Use Framework for State-of-the-Art NLP. In Proc. of the 2019 Conference of the North American Chapter
of the Association for Computational Linguistics (Demonstrations) . ACL, Minneapolis, Minnesota, 54–59. https:
//doi.org/10.18653/v1/N19-4010
[4]Alan Akbik, Duncan Blythe, and Roland Vollgraf. 2018. Contextual String Embeddings for Sequence Labeling. In
Proc. of the 27th International Conference on Computational Linguistics . ACL, Santa Fe, New Mexico, USA, 1638–1649.
https://www.aclweb.org/anthology/C18-113940 Ehrmann et al.
[5]Beatrice Alex, Kate Byrne, Claire Grover, and R. Tobin. 2015. Adapting the Edinburgh Geoparser for Historical
Georeferencing. International Journal of Humanities and Arts Computing 9, 1 (2015), 15–35. https://doi.org/10.3366/
ijhac.2015.0136
[6]Bea Alex, Claire Grover, Ewan Klein, and Richard Tobin. 2012. Digitised Historical Text: Does It Have to Be mediOCRe?.
InProc. of KONVENS 2012 , Jeremy Jancsary (Ed.). ÖGAI, Vienna, Austria, 401–409. http://www.oegai.at/konvens2012/
proceedings/59\_alex12w/
[7]Beatrice Alex, Claire Grover, R. Tobin, and J. Oberlander. 2019. Geoparsing Historical and Contemporary Literary
Text Set in the City of Edinburgh. Lang Resources & Evaluation 53 (2019), 651–675. https://doi.org/10.1007/S10579-
019-09443-X
[8] Isabelle Augenstein, Leon Derczynski, and Kalina Bontcheva. 2017. Generalisation in Named Entity Recognition: A
Quantitative Analysis. Computer Speech & Language 44 (July 2017), 61–83. https://doi.org/10.1016/j.csl.2017.01.012
[9]Alexandra Balahur, Ralf Steinberger, Mijail A. Kabadjov, Vanni Zavarella, Erik Van der Goot, Matina Halkia, Bruno
Pouliquen, and Jenya Belyaeva. 2010. Sentiment Analysis in the News. In Proc. of the International Conference
on Language Resources and Evaluation (LREC’10) , Nicoletta Calzolari, Khalid Choukri, Bente Maegaard, Joseph
Mariani, Jan Odijk, Stelios Piperidis, Mike Rosner, and Daniel Tapias (Eds.). ELRA, Valletta, Malta, 2216–2220.
http://www.lrec-conf.org/proceedings/lrec2010/summaries/909.html
[10] Timothy Baldwin, Paul Cook, Marco Lui, Andrew MacKinlay, and Li Wang. 2013. How Noisy Social Media Text, How
Diffrnt Social Media Sources?. In Proc. of the Sixth International Joint Conference on Natural Language Processing . Asian
Federation of Natural Language Processing, Nagoya, Japan, 356–364. https://www.aclweb.org/anthology/I13-1041
[11] David Bamman and Patrick J. Burns. 2020. Latin BERT: A Contextual Language Model for Classical Philology. (Sept.
2020). arXiv:2009.10053 http://arxiv.org/abs/2009.10053
[12] David Bamman, Sejal Popat, and Sheng Shen. 2019. An Annotated Dataset of Literary Entities. In Proc. of the
2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language
Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and Short Papers) , Jill Burstein,
Christy Doran, and Thamar Solorio (Eds.). ACL, Minneapolis, Minnesota, 2138–2144. https://doi.org/10.18653/v1/n19-
1220
[13] Marco Baroni and Alessandro Lenci. 2010. Distributional Memory: A General Framework for Corpus-Based Semantics.
Computational Linguistics 36, 4 (Oct. 2010), 673–721. https://doi.org/10.1162/coli_a_00016
[14] Marcia J. Bates. 1996. The Getty End-User Online Searching Project in the Humanities: Report No. 6: Overview
and Conclusions | Bates | College & Research Libraries. College & Research Libraries 57, 6 (1996), 514–523. https:
//doi.org/10.5860/crl_57_06_514
[15] Oliver Bender, Franz Josef Och, and Hermann Ney. 2003. Maximum Entropy Models for Named Entity Recognition.
InProc. of the Seventh Conference on Natural Language Learning at HLT-NAACL 2003 - Volume 4 (CONLL ’03) . ACL,
USA, 148–151. https://doi.org/10.3115/1119176.1119196
[16] Y. Bengio, A. Courville, and P. Vincent. 2013. Representation Learning: A Review and New Perspectives. IEEE
Transactions on Pattern Analysis and Machine Intelligence 35, 8 (Aug. 2013), 1798–1828. https://doi.org/10.1109/TPAMI.
2013.50
[17] Yoshua Bengio, Réjean Ducharme, Pascal Vincent, and Christian Jauvin. 2003. A Neural Probabilistic Language Model.
Journal of Machine Learning Research 3, Feb (2003), 1137–1155. https://www.jmlr.org/papers/v3/bengio03a.html
[18] Darina Benikova, Chris Biemann, and Marc Reznicek. 2014. NoSta-D Named Entity Annotation for German: Guidelines
and Dataset.. In Proc. of the Ninth International Conference on Language Resources and Evaluation (LREC’14) . ELRA,
Reykjavik, Iceland, 2524–2531.
[19] Daniel M. Bikel, Scott Miller, Richard Schwartz, and Ralph Weischedel. 1997. Nymble: A High-Performance Learning
Name-Finder. In Fifth Conference on Applied Natural Language Processing . ACL, Washington, DC, USA, 194–201.
https://doi.org/10.3115/974557.974586
[20] Stefan Bircher. 2019. Toulouse and Cahors Are French Cities, but T «i* Louse and Caa. Qrs as Well . PhD Thesis. University
of Zurich.
[21] Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2017. Enriching Word Vectors with Subword
Information. Transactions of the Association for Computational Linguistics 5 (2017), 135–146. https://www.aclweb.
org/anthology/Q17-1010
[22] Kalina Bontcheva, Diana Maynard, Hamish Cunningham, and Horacio Saggion. 2002. Using Human Language
Technology for Automatic Annotation and Indexing of Digital Library Content. In Research and Advanced Technology
for Digital Libraries (Lecture Notes in Computer Science) , Maristella Agosti and Costantino Thanos (Eds.). Springer,
Berlin, Heidelberg, 613–625. https://doi.org/10.1007/3-540-45747-X_46
[23] Lars Borin, Dimitrios Kokkinakis, and Leif-Jöran Olsson. 2007. Naming the Past: Named Entity and Animacy
Recognition in 19th Century Swedish Literature. In Proc. of the Workshop on Language Technology for Cultural Heritage
Data (LaTeCH 2007). ACL, Prague, Czech Republic, 1–8. https://www.aclweb.org/anthology/W07-0901Named Entity Recognition and Classification on Historical Documents: A Survey 41
[24] Emanuela Boros, Ahmed Hamdi, Elvys Linhares Pontes, Luis Adrián Cabrera-Diego, Jose G. Moreno, Nicolas Sidere,
and Antoine Doucet. 2020. Alleviating Digitization Errors in Named Entity Recognition for Historical Documents.
InProc. of the 24th Conference on Computational Natural Language Learning . ACL, Online, 431–441. https://www.
aclweb.org/anthology/2020.conll-1.35
[25] Emanuela Boros, Elvys Linhares Pontes, Luis Adrián Cabrera-Diego, Ahmed Hamdi, Jose G. Moreno, Nicolas Sidère,
and Antoine Doucet. 2020. Robust Named Entity Recognition and Linking on Historical Multilingual Documents.
InWorking Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda Cappellato, Carsten Eickhoff,
Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 1–17. http://ceur-ws.org/Vol-
2696/paper_171.pdf
[26] Frank Bösch. 2019. Zeitenwende 1979. Als Die Welt von Heute Begann . C.H. Beck Verlag, Munich.
[27] Federico Boschetti, Stefano Dei Rossi, Felice Dell’Orletta, Michele Di Giorgio, Martina Miliani, Lucia C. Passaro,
Angelica Puddu, Giulia Venturi, Nicola Labanca, and Alessandro Lenci. 2020. “Voices of the Great War”: A Richly
Annotated Corpus of Italian Texts on the First World War. In Proc. of The 12th Language Resources and Evaluation
Conference . ELRA, Marseille, France, 911–918. https://www.aclweb.org/anthology/2020.lrec-1.114
[28] Yanne Broux and Mark Depauw. 2015. Developing Onomastic Gazetteers and Prosopographies for the Ancient World
through Named Entity Recognition and Graph Visualization: Some Examples from Trismegistos People. In Social
Informatics , Luca Maria Aiello and Daniel McFarland (Eds.). Springer International Publishing, Cham, 304–313.
[29] David Campos, Sérgio Matos, and José Luís Oliveira. 2012. Biomedical Named Entity Recognition: A Survey of
Machine-Learning Tools. In Theory and Applications for Advanced Text Mining , Shigeaki Sakurai (Ed.). IntechOpen,
Rijeka, 175–195. https://doi.org/10.5772/51066
[30] Steve Cassidy. 2016. Publishing the Trove Newspaper Corpus. In Proc. of the Tenth International Conference on Language
Resources and Evaluation (LREC’16) . ELRA, Portorož, Slovenia, 4520–4525. https://www.aclweb.org/anthology/L16-
1715
[31] Tim Causer and Melissa Terras. 2014. “Many Hands Make Light Work. Many Hands Together Make Merry Work”:
Transcribe Bentham and Crowdsourcing Manuscript Collections. In Crowdsourcing Our Cultural Heritage . Ashgate,
Farnham, 57–88. http://www.ashgate.com/isbn/9781472410221
[32] Anne Chardonnens, Ettore Rizza, Mathias Coeckelbergs, and Seth van Hooland. 2017. Mining User Queries with
Information Extraction Methods and Linked Data. Journal of Documentation 74, 5 (2017), 936–950. https://doi.org/10.
1108/JD-09-2017-0133
[33] Guillaume Chiron, Antoine Doucet, Mickaël Coustaty, Muriel Visani, and Jean-Philippe Moreux. 2017. Impact of
OCR Errors on the Use of Digital Libraries: Towards a Better Access to Information. In Proc. of the 17th ACM/IEEE
Joint Conference on Digital Libraries . IEEE, IEEE Press, Toronto, ON, Canada, 249–252. https://doi.org/10.1109/JCDL.
2017.7991582
[34] Kyunghyun Cho, Bart van Merriënboer, Caglar Gulcehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk,
and Yoshua Bengio. 2014. Learning Phrase Representations Using RNN Encoder–Decoder for Statistical Machine
Translation. In Proc. of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) . ACL, Doha,
Qatar, 1724–1734. https://doi.org/10.3115/v1/D14-1179
[35] Giovanni Colavizza, Maud Ehrmann, and Fabio Bortoluzzi. 2019. Index-Driven Digitization and Indexation of
Historical Archives. Frontiers in Digital Humanities 6 (2019), 4. https://doi.org/10.3389/fdigh.2019.00004
[36] Giovanni Colavizza and Matteo Romanello. 2017. Annotated References in the Historiography on Venice: 19th–21st
Centuries. Journal of Open Humanities Data 3, 0 (Nov. 2017), 2. https://doi.org/10.5334/johd.9
[37] Ronan Collobert, Jason Weston, Leon Bottou, Michael Karlen, Koray Kavukcuoglu, and Pavel Kuksa. 2011. Natural
Language Processing (Almost) from Scratch. Journal of machine learning research 12 (2011), 2493–2537.
[38] Hamish Cunningham, Diana Maynard, Kalina Bontcheva, and Valentin Tablan. 2002. GATE: A Framework and
Graphical Development Environment for Robust NLP Tools and Applications. In Proc. 40th Annual Meeting of
the Association for Computational Linguistics (ACL 2002) . ACL, Philadelphia, Pennsylvania, USA, 168–175. https:
//openreview.net/forum?id=HkZEKhgubS
[39] Xiang Dai and Heike Adel. 2020. An Analysis of Simple Data Augmentation for Named Entity Recognition. In Proc. of
the 28th International Conference on Computational Linguistics . International Committee on Computational Linguistics,
Barcelona, Spain (Online), 3861–3867. https://www.aclweb.org/anthology/2020.coling-main.343
[40] Mark Davies. 2012. Expanding Horizons in Historical Linguistics with the 400-Million Word Corpus of Historical
American English. Corpora 7, 2 (2012), 121–157.
[41] Ghaith Dekhili and Fatiha Sadat. 2020. Hybrid Statistical and Attentive Deep Neural Approach for Named Entity
Recognition in Historical Newspapers. In Working Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum ,
Linda Cappellato, Carsten Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece,
1–17. http://ceur-ws.org/Vol-2696/paper_199.pdf42 Ehrmann et al.
[42] Leon Derczynski, Eric Nichols, Marieke van Erp, and Nut Limsopatham. 2017. Results of the WNUT2017 Shared
Task on Novel and Emerging Entity Recognition. In Proc. of the 3rd Workshop on Noisy User-Generated Text . ACL,
Copenhagen, Denmark, 140–147. https://doi.org/10.18653/v1/W17-4418
[43] Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-Training of Deep Bidirectional
Transformers for Language Understanding. In Proc. of the 2019 Conference of the North American Chapter of the
Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) . ACL,
Minneapolis, Minnesota, 4171–4186. https://doi.org/10.18653/v1/N19-1423
[44] Marco Dinarelli and Sophie Rosset. 2012. Tree-Structured Named Entity Recognition on OCR Data: Analysis,
Processing and Results. In Proc. of the Eight International Conference on Language Resources and Evaluation (LREC’12) ,
Nicoletta Calzolari (Conference Chair), Khalid Choukri, Thierry Declerck, Mehmet Ugur Dogan, Bente Maegaard,
Joseph Mariani, Jan Odijk, and Stelios Piperidis (Eds.). ELRA, Istanbul, Turkey, 23–25.
[45] George Doddington, Alexis Mitchell, Mark Przybocki, Lance Ramshaw, Stephanie Strassel, and Ralph Weischedel. 2004.
The Automatic Content Extraction (ACE) Program – Tasks, Data, and Evaluation. In Proc. of the Fourth International
Conference on Language Resources and Evaluation (LREC’04) . ELRA, Lisbon, Portugal, 837–840. http://www.lrec-
conf.org/proceedings/lrec2004/pdf/5.pdf
[46] Jad Doughman, Fatima Abu Salem, and Shady Elbassuoni. 2020. Time-Aware Word Embeddings for Three Lebanese
News Archives. In Proc. of the 12th Language Resources and Evaluation Conference . ELRA, Marseille, France, 4717–4725.
https://www.aclweb.org/anthology/2020.lrec-1.580
[47] Wendy M. Duff and Catherine A. Johnson. 2002. Accidentally Found on Purpose: Information-Seeking Behavior of
Historians in Archives. The Library Quarterly 72, 4 (2002), 472–496.
[48] Maud Ehrmann. 2008. Named Entities, from Linguistics to NLP: Theoretical Status and Disambiguation Methods (Les
Entitées Nommées, de La Linguistique Au TAL : Statut Théorique et Méthodes de Désambiguïsation) . Theses. Paris
Diderot University. https://hal.archives-ouvertes.fr/tel-01639190
[49] Maud Ehrmann, Estelle Bunout, and Marten Düring. 2019. Historical Newspaper User Interfaces: A Review. In Proc. of
the 85thIFLA General Conference and Assembly . IFLA Library, Athens, Greece, 1–26. https://doi.org/10.5281/zenodo.
3404155
[50] Maud Ehrmann, Giovanni Colavizza, Yannick Rochat, and Frédéric Kaplan. 2016. Diachronic Evaluation of NER
Systems on Old Newspapers. In Proc. of the 13th Conference on Natural Language Processing (KONVENS 2016) . Bochumer
Linguistische Arbeitsberichte, Bochum, 97–107. https://infoscience.epfl.ch/record/221391
[51] Maud Ehrmann, Damien Nouvel, and Sophie Rosset. 2016. Named Entity Resources - Overview and Outlook. In Proc. of
the Tenth International Conference on Language Resources and Evaluation (LREC’16) , Nicoletta Calzolari, Khalid Choukri,
Thierry Declerck, Sara Goggi, Marko Grobelnik, Bente Maegaard, Joseph Mariani, Helene Mazo, Asuncion Moreno,
Jan Odijk, and Stelios Piperidis (Eds.). ELRA, Paris, France, 3349–3356. https://www.aclweb.org/anthology/L16-1534
[52] Maud Ehrmann, Matteo Romanello, Simon Clematide, Phillip Benjamin Ströbel, and Raphaël Barman. 2020. Language
Resources for Historical Newspapers: The Impresso Collection. In Proc. of the 12th Language Resources and Evaluation
Conference . ELRA, Marseille, France, 958–968. https://www.aclweb.org/anthology/2020.lrec-1.121
[53] Maud Ehrmann, Matteo Romanello, Alex Flückiger, and Simon Clematide. 2020. Extended Overview of CLEF HIPE
2020: Named Entity Processing on Historical Newspapers. In Working Notes of CLEF 2020 - Conference and Labs of the
Evaluation Forum , Linda Cappellato, Carsten Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS,
Thessaloniki, Greece, 38. https://doi.org/10.5281/zenodo.4117566
[54] Maud Ehrmann, Matteo Romanello, Alex Flückiger, and Simon Clematide. 2020. Impresso Named Entity Annotation
Guidelines . Annotation Guidelines. Ecole Polytechnique Fédérale de Lausanne (EPFL) and Zurich University (UZH).
https://doi.org/10.5281/zenodo.3604227
[55] Maud Ehrmann, Matteo Romanello, Alex Flückiger, and Simon Clematide. 2020. Overview of CLEF HIPE 2020: Named
Entity Recognition and Linking on Historical Newspapers. In Experimental IR Meets Multilinguality, Multimodality,
and Interaction (Lecture Notes in Computer Science) , Avi Arampatzis, Evangelos Kanoulas, Theodora Tsikrika, Stefanos
Vrochidis, Hideo Joho, Christina Lioma, Carsten Eickhoff, Aurélie Névéol, Linda Cappellato, and Nicola Ferro (Eds.).
Springer International Publishing, Cham, 288–310. https://doi.org/10.1007/978-3-030-58219-7_21
[56] Jacob Eisenstein. 2013. What to Do about Bad Language on the Internet. In Proc. of the 2013 Conference of the North
American Chapter of the Association for Computational Linguistics: Human Language Technologies . ACL, Atlanta,
Georgia, 359–369. https://www.aclweb.org/anthology/N13-1037
[57] Asif Ekbal, Eva Sourjikova, Anette Frank, and Simone Paolo Ponzetto. 2010. Assessing the Challenge of Fine-Grained
Named Entity Recognition and Classification. In Proc. of the 2010 Named Entities Workshop (NEWS ’10) . ACL, USA,
93–101.
[58] Cheikh Brahim El Vaigh, Guillaume Le Noé-Bienvenu, Guillaume Gravier, and Pascale Sébillot. 2020. IRISA System
for Entity Detection and Linking at HIPE’20. In Working Notes of CLEF 2020 - Conference and Labs of the Evaluation
Forum , Linda Cappellato, Carsten Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki,Named Entity Recognition and Classification on Historical Documents: A Survey 43
Greece, 1–7. http://ceur-ws.org/Vol-2696/paper_185.pdf
[59] Jeffrey L. Elman. 1990. Finding Structure in Time. Cognitive Science 14, 2 (1990), 179–211. https://doi.org/10.1207/
s15516709cog1402_1
[60] Alex Erdmann, Christopher Brown, Brian D. Joseph, Mark Janse, Petra Ajaka, Micha Elsner, and Marie-Catherine de
Marneffe. 2016. Challenges and Solutions for Latin Named Entity Recognition. In COLING 2016: 26th International
Conference on Computational Linguistics . ACL, Osaka, Japan, 85–93.
[61] Alexander Erdmann, David Joseph Wrisley, Benjamin Allen, Christopher Brown, Sophie Cohen-Bodénès, Micha Elsner,
Yukun Feng, Brian Joseph, Béatrice Joyeux-Prunel, and Marie-Catherine de Marneffe. 2019. Practical, Efficient, and
Customizable Active Learning for Named Entity Recognition in the Digital Humanities. In Proc. of the 2019 Conference
of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume
1 (Long and Short Papers) . ACL, Minneapolis, Minnesota, 2223–2234. https://doi.org/10.18653/v1/N19-1231
[62] Llorenç Escoter, Lidia Pivovarova, Mian Du, Anisia Katinskaia, and Roman Yangarber. 2017. Grouping Business
News Stories Based on Salience of Named Entities. In Proc. of the 15th Conference of the European Chapter of
the Association for Computational Linguistics: Volume 1, Long Papers . ACL, Valencia, Spain, 1096–1106. https:
//www.aclweb.org/anthology/E17-1103
[63] Jenny Rose Finkel, Trond Grenager, and Christopher Manning. 2005. Incorporating Non-Local Information into
Information Extraction Systems by Gibbs Sampling. In Proc. of the 43rd Annual Meeting on Association for Computational
Linguistics . ACL, Ann Arbor, Michigan, 363–370.
[64] Antske Fokkens, Serge ter Braake, Ronald Sluijter, Paul Longley Arthur, and Eveline Wandl-Vogt (Eds.). 2018. Proc. of
the Second Conference on Biographical Data in a Digital World 2017 . CEUR Workshop Proc., Vol. 2119. CEUR-WS.org,
Linz, Austria. http://ceur-ws.org/Vol-2119
[65] Hege Fromreide, Dirk Hovy, and Anders Søgaard. 2014. Crowdsourcing and Annotating NER for Twitter #drift. In
Proc. of the Ninth International Conference on Language Resources and Evaluation (LREC’14) . ELRA, Reykjavik, Iceland,
2544–2547. http://www.lrec-conf.org/proceedings/lrec2014/pdf/421_Paper.pdf
[66] Olivier Galibert, Jeremy Leixa, Gilles Adda, Khalid Choukri, and Guillaume Gravier. 2014. The ETAPE Speech
Processing Evaluation. In Proc. of the Ninth International Conference on Language Resources and Evaluation (LREC’14) .
ELRA, Reykjavik, Iceland, 3995–3999. http://www.lrec-conf.org/proceedings/lrec2014/pdf/1027_Paper.pdf
[67] O. Galibert, S. Rosset, C. Grouin, P. Zweigenbaum, and L. Quintard. 2012. Extended Named Entity Annotation
on OCRed Documents : From Corpus Constitution to Evaluation Campaign. In Proc. of the Eighth Conference on
International Language Resources and Evaluation . ELRA, Istanbul, Turkey, 3126–3131.
[68] Tiberiu-Marian Georgescu, Bogdan Iancu, Alin Zamfiroiu, Mihai Doinea, Catalin Emilian Boja, and Cosmin Cartas.
2021. A Survey on Named Entity Recognition Solutions Applied for Cybersecurity-Related Text Processing. In
Proc. of Fifth International Congress on Information and Communication Technology (Advances in Intelligent Systems
and Computing) , Xin-She Yang, Simon Sherratt, Nilanjan Dey, and Amit Joshi (Eds.). Springer, Singapore, 316–325.
https://doi.org/10.1007/978-981-15-5859-7_31
[69] Fredric C. Gey. 2000. Research to Improve Cross-Language Retrieval - Position Paper for CLEF. In Revised Papers from
the Workshop of Cross-Language Evaluation Forum on Cross-Language Information Retrieval and Evaluation (CLEF ’00) .
Springer-Verlag, Berlin, Heidelberg, 83–88.
[70] Sahar Ghannay, Cyril Grouin, and Thomas Lavergne. 2020. Experiments from LIMSI at the French Named Entity
Recognition Coarse-Grained Task. In Working Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda
Cappellato, Carsten Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 1–8.
http://ceur-ws.org/Vol-2696/paper_152.pdf
[71] Paul Gooding. 2016. Exploring the Information Behaviour of Users of Welsh Newspapers Online through Web Log
Analysis. Journal of Documentation 72, 2 (Jan. 2016), 232–246. https://doi.org/10.1108/JD-10-2014-0149
[72] Sergiu Gordea, Monica Lestari Paramita, and Antoine Isaac. 2020. Named Entity Recommendations to Enhance
Multilingual Retrieval in Europeana.Eu. In Foundations of Intelligent Systems (Lecture Notes in Computer Science) , Denis
Helic, Gerhard Leitner, Martin Stettinger, Alexander Felfernig, and Zbigniew W. Raś (Eds.). Springer International
Publishing, Cham, 102–112. https://doi.org/10.1007/978-3-030-59491-6_10
[73] Rodrigo Rafael Villarreal Goulart, Vera Lúcia Strube de Lima, and Clarissa Castellã Xavier. 2011. A Systematic Review
of Named Entity Recognition in Biomedical Texts. Journal of the Brazilian Computer Society 17, 2 (June 2011), 103–116.
https://doi.org/10.1007/s13173-011-0031-9
[74] Sara Grilo, Márcia Bolrinha, João Silva, Rui Vaz, and António Branco. 2020. The BDCamões Collection of Portuguese
Literary Documents: A Research Resource for Digital Humanities and Language Technology. In Proc. of the 12th
Language Resources and Evaluation Conference . ELRA, Marseille, France, 849–854. https://www.aclweb.org/anthology/
2020.lrec-1.106
[75] Ralph Grishman and Beth Sundheim. 1995. Design of the MUC-6 Evaluation. In Proc. of the 6th Conference on Message
Understanding (MUC6 ’95) . ACL, USA, 1–11. https://doi.org/10.3115/1072399.107240144 Ehrmann et al.
[76] Ralph Grishman and Beth Sundheim. 1996. Message Understanding Conference- 6: A Brief History. In COLING 1996
Volume 1: The 16th International Conference on Computational Linguistics . Center for Sprogteknologi, Copenhagen,
Denmark, 466–471. https://www.aclweb.org/anthology/C96-1079
[77] Claire Grover, Sharon Givon, Richard Tobin, and Julian Ball. 2008. Named Entity Recognition for Digitised Historical
Texts. In Proc. of the Sixth International Conference on Language Resources and Evaluation (LREC’08) . ELRA, Marrakech,
Morocco, 1343–1346. http://www.lrec-conf.org/proceedings/lrec2008/pdf/342_paper.pdf
[78] Jiafeng Guo, Gu Xu, Xueqi Cheng, and Hang Li. 2009. Named Entity Recognition in Query. In Proc. of the 32nd
International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR ’09) . Association for
Computing Machinery, New York, NY, USA, 267–274. https://doi.org/10.1145/1571941.1571989
[79] Ahmed Hamdi, Axel Jean-Caurant, Nicolas Sidere, Mickaël Coustaty, and Antoine Doucet. 2019. An Analysis of the
Performance of Named Entity Recognition over OCRed Documents. In Proc. of the 18th Joint Conference on Digital
Libraries (JCDL ’19) . IEEE, IEEE Press, Champaign, Illinois, 333–334. https://doi.org/10.1109/JCDL.2019.00057
[80] Ahmed Hamdi, Axel Jean-Caurant, Nicolas Sidère, Mickaël Coustaty, and Antoine Doucet. 2020. Assessing and
Minimizing the Impact of OCR Quality on Named Entity Recognition. In Digital Libraries for Open Knowledge
(Lecture Notes in Computer Science) , Mark Hall, Tanja Merčun, Thomas Risse, and Fabien Duchateau (Eds.). Springer
International Publishing, Cham, 87–101. https://doi.org/10.1007/978-3-030-54956-5_7
[81] Ahmed Hamdi, Elvys Linhares Pontes, Emanuela Boros, Thi Tuyet Hai Nguyen, Günter Hackl, Jose G. Moreno,
and Antoine Doucet. 2021. A Multilingual Dataset for Named Entity Recognition, Entity Linking and Stance
Detection in Historical Newspapers. In SIGIR ’21: Proc. of the 43rd International ACM SIGIR Conference on Research and
Development in Information Retrieval . Association for Computing MachineryNew YorkNYUnited States, Montreal,
Canada. https://doi.org/10.5281/zenodo.4694466
[82] William L. Hamilton, Jure Leskovec, and Dan Jurafsky. 2016. Diachronic Word Embeddings Reveal Statistical Laws of
Semantic Change. In Proc. of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long
Papers) . ACL, Berlin, Germany, 1489–1501. https://doi.org/10.18653/v1/P16-1141
[83] Simon Hengchen, Ruben Ros, Jani Marjanen, and Mikko Tolonen. 2019. Models for "A Data-Driven Approach to
Studying Changing Vocabularies in Historical Newspaper Collections". NewsEye Project. https://doi.org/10.5281/
zenodo.3585027
[84] Simon Hengchen and Nina Tahmasebi. 2021. A Collection of Swedish Diachronic Word Embedding Models Trained
on Historical Newspaper Data. Journal of Open Humanities Data 7, 0 (Jan. 2021), 2. https://doi.org/10.5334/johd.22
[85] Ulf Hermjakob, Kevin Knight, and Hal Daumé III. 2008. Name Translation in Statistical Machine Translation - Learning
When to Transliterate. In Proc. of ACL-08: HLT . ACL, Columbus, Ohio, 389–397. https://www.aclweb.org/anthology/
P08-1045
[86] Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long Short-Term Memory. Neural Computation 9, 8 (Nov. 1997),
1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735
[87] Kasra Hosseini, Kaspar Beelen, Giovanni Colavizza, and Mariona Coll Ardanuy. 2021. Neural Language Models for
Nineteenth-Century English. arXiv:2105.11321 [cs] (May 2021), 5. arXiv:2105.11321 [cs] http://arxiv.org/abs/2105.11321
[88] Jeremy Howard and Sebastian Ruder. 2018. Universal Language Model Fine-Tuning for Text Classification. In Proc. of
the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) . ACL, Melbourne,
Australia, 328–339. https://doi.org/10.18653/v1/P18-1031
[89] Zhiheng Huang, Wei Xu, and Kai Yu. 2015. Bidirectional LSTM-CRF Models for Sequence Tagging. (2015). https:
//arxiv.org/abs/1508.01991
[90] Helena Hubková. 2019. Named-Entity Recognition in Czech Historical Texts: Using a CNN-BiLSTM Neural Network
Model . Ph.D. Dissertation. Uppsala University.
[91] Helena Hubková, Pavel Kral, and Eva Pettersson. 2020. Czech Historical Named Entity Corpus v 1.0. In Proc. of the
12th Language Resources and Evaluation Conference . ELRA, Marseille, France, 4458–4465. https://www.aclweb.org/
anthology/2020.lrec-1.549
[92] Bouke Huurnink, Laura Hollink, Wietske van den Heuvel, and Maarten de Rijke. 2010. Search Behavior of Media
Professionals at an Audiovisual Archive: A Transaction Log Analysis. Journal of the American Society for Information
Science and Technology 61, 6 (2010), 1180–1197. https://doi.org/10.1002/asi.21327
[93] Vinh-Nam Huynh, Ahmed Hamdi, and Antoine Doucet. 2020. When to Use OCR Post-Correction for Named Entity
Recognition?. In Digital Libraries at Times of Massive Societal Transition (Lecture Notes in Computer Science) , Emi
Ishita, Natalie Lee San Pang, and Lihong Zhou (Eds.). Springer International Publishing, Cham, 33–42. https:
//doi.org/10.1007/978-3-030-64452-9_3
[94] Hideki Isozaki and Hideto Kazawa. 2002. Efficient Support Vector Classifiers for Named Entity Recognition. In
COLING 2002: The 19th International Conference on Computational Linguistics (COLING ’02) . ACL, Taipei, Taiwan, 1–7.
https://doi.org/10.3115/1072228.1072282Named Entity Recognition and Classification on Historical Documents: A Survey 45
[95] A. Jones and G. Crane. 2006. The Challenge of Virginia Banks: An Evaluation of NamedEntity Analysis in a 19th-
Century Newspaper Collection. In Proc. of the 6th ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL ’06) . IEEE,
Chapel Hill, NC, USA, 31–40. https://doi.org/10.1145/1141753.1141759
[96] Pratik Joshi, Sebastin Santy, Amar Budhiraja, Kalika Bali, and Monojit Choudhury. 2020. The State and Fate of
Linguistic Diversity and Inclusion in the NLP World. In Proc. of the 58th Annual Meeting of the Association for
Computational Linguistics . ACL, Online, 6282–6293. https://doi.org/10.18653/v1/2020.acl-main.560
[97] Mijail Kabadjov, Josef Steinberger, and Ralf Steinberger. 2013. Multilingual Statistical News Summarization. In Multi-
Source, Multilingual Information Extraction and Summarization , Thierry Poibeau, Horacio Saggion, Jakub Piskorski,
and Roman Yangarber (Eds.). Springer, Berlin, Heidelberg, 229–252. https://doi.org/10.1007/978-3-642-28569-1_11
[98] Frédéric Kaplan and Isabella di Lenardo. 2017. Big Data of the Past. Frontiers in Digital Humanities 4 (2017), 1–21.
https://doi.org/10.3389/fdigh.2017.00012
[99] Kimmo Kettunen, Timo Honkela, Krister Lindén, Pekka Kauppinen, Tuula Pääkkönen, Jukka Kervinen, et al .2014.
Analyzing and Improving the Quality of a Historical News Collection Using Language Technology and Statistical
Machine Learning Methods. In IFLA World Library and Information Congress Proc. 80th IFLA General Conference and
Assembly . IFLA, Lyon, France, 1–23.
[100] Kimmo Kettunen and Teemu Ruokolainen. 2017. Names, Right or Wrong: Named Entities in an OCRed Historical
Finnish Newspaper Collection. In Proc. of the 2nd International Conference on Digital Access to Textual Cultural Heritage
- DATeCH2017 . ACM Press, G&#246;ttingen, Germany, 181–186. https://doi.org/10.1145/3078081.3078084
[101] Tannon Kew, Anastassia Shaitarova, Isabel Meraner, Janis Goldzycher, Simon Clematide, and Martin Volk. 2019.
Geotagging a Diachronic Corpus of Alpine Texts: Comparing Distinct Approaches to Toponym Recognition. In
Proc. of the Workshop on Language Technology for Digital Historical Archives . INCOMA Ltd., Varna, Bulgaria, 11–18.
https://doi.org/10.26615/978-954-452-059-5_003
[102] J.-D. Kim, T. Ohta, Y. Tateisi, and J. Tsujii. 2003. GENIA Corpus—a Semantically Annotated Corpus for Bio-Textmining.
Bioinformatics 19, suppl_1 (July 2003), i180–i182. https://doi.org/10.1093/bioinformatics/btg1023
[103] Sunghwan Mac Kim and Steve Cassidy. 2015. Finding Names in Trove: Named Entity Recognition for Australian
Historical Newspapers. In Proc. of the Australasian Language Technology Association Workshop 2015 . ACL, Parramatta,
Australia, 57–65. https://www.aclweb.org/anthology/U15-1007
[104] Eleni Kogkitsidou and Philippe Gambette. 2020. Normalisation of 16th and 17th Century Texts in French and
Geographical Named Entity Recognition. In Proc. of the 4th ACM SIGSPATIAL Workshop on Geospatial Humanities
(GeoHumanities’20) . Association for Computing Machinery, New York, NY, USA, 28–34. https://doi.org/10.1145/
3423337.3429437
[105] Tanti Kristanti and Laurent Romary. 2020. DeLFT and Entity-Fishing: Tools for CLEF HIPE 2020 Shared Task. In
Working Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda Cappellato, Carsten Eickhoff,
Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 1–10. http://ceur-ws.org/Vol-
2696/paper_266.pdf
[106] Markus Krug, Lukas Weimer, Isabella Reger, Luisa Macharowsky, Stephan Feldhaus, Frank Puppe, and Fotis Jannidis.
2018. Description of a Corpus of Character References in German Novels - DROC [Deutsches ROman Corpus] . Technical
Report. SUB Göttingen. http://nbn-resolving.org/urn:nbn:de:gbv:7-dariah-2018-2-9
[107] Kai Labusch and Clemens Neudecker. 2020. Named Entity Disambiguation and Linking Historic Newspaper OCR
with BERT. In Working Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda Cappellato, Carsten
Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 1–14. http://ceur-
ws.org/Vol-2696/paper_163.pdf
[108] Kai Labusch, Clemens Neudecker, and David Zellhöfer. 2019. BERT for Named Entity Recognition in Contemporary
and Historic German. In Preliminary Proc. of the 15th Conference on Natural Language Processing (KONVENS 2019) .
German Society for Computational Linguistics & Language Technology, Erlangen, Germany, 1–9.
[109] John D. Lafferty, Andrew McCallum, and Fernando C. N. Pereira. 2001. Conditional Random Fields: Probabilistic
Models for Segmenting and Labeling Sequence Data. In Proc. of the Eighteenth International Conference on Machine
Learning (ICML ’01) . Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 282–289.
[110] Guillaume Lample, Miguel Ballesteros, Sandeep Subramanian, Kazuya Kawakami, and Chris Dyer. 2016. Neural
Architectures for Named Entity Recognition. In Proc. of the 2016 Conference of the North American Chapter of the
Association for Computational Linguistics: Human Language Technologies . ACL, San Diego, California, 260–270.
https://doi.org/10.18653/v1/N16-1030
[111] Robert Leaman and Graciela Gonzalez. 2008. BANNER: An Executable Survey of Advances in Biomedical Named
Entity Recognition. In Proc. of the Pacific Symposium on Biocomputing 2008 . World Scientific Publishing, Kohala Coast,
Hawaii, USA, 652–663. https://doi.org/10.1142/6646
[112] Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. 2015. Deep Learning. Nature 521, 7553 (May 2015), 436–444.
https://doi.org/10.1038/nature1453946 Ehrmann et al.
[113] Jens Lehmann, Robert Isele, Max Jakob, Anja Jentzsch, Dimitris Kontokostas, Pablo N. Mendes, Sebastian Hellmann,
Mohamed Morsey, Patrick van Kleef, Sören Auer, and Christian Bizer. 2015. DBpedia - A Large-Scale, Multilingual
Knowledge Base Extracted from Wikipedia. Semantic Web Journal 6, 2 (2015), 167–195. http://jens-lehmann.org/
files/2015/swj_dbpedia.pdf
[114] J. Lei, Buzhou Tang, Xueqin Lu, Kaihua Gao, Min Jiang, and H. Xu. 2014. Research and Applications: A Comprehensive
Study of Named Entity Recognition in Chinese Clinical Text. Journal of the American Medical Informatics Association
21, 5 (2014), 808–814. https://doi.org/10.1136/amiajnl-2013-002381
[115] Omer Levy and Yoav Goldberg. 2014. Dependency-Based Word Embeddings. In Proc. of the 52nd Annual Meeting of
the Association for Computational Linguistics (Volume 2: Short Papers) . ACL, Baltimore, Maryland, 302–308. https:
//doi.org/10.3115/v1/P14-2050
[116] Jing Li, Aixin Sun, Jianglei Han, and Chenliang Li. 2020. A Survey on Deep Learning for Named Entity Recognition.
IEEE Transactions on Knowledge and Data Engineering (2020), 1–20. https://doi.org/10.1109/TKDE.2020.2981314
[117] Hongyu Lin, Yaojie Lu, Jialong Tang, Xianpei Han, Le Sun, Zhicheng Wei, and Nicholas Jing Yuan. 2020. A Rigorous
Study on Named Entity Recognition: Can Fine-Tuning Pretrained Model Lead to the Promised Land?. In Proc.
of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) . ACL, Online, 7291–7300.
https://www.aclweb.org/anthology/2020.emnlp-main.592
[118] Thomas Lin, Patrick Pantel, Michael Gamon, Anitha Kannan, and Ariel Fuxman. 2012. Active Objects: Actions for
Entity-Centric Search. In Proc. of the 21st International Conference on World Wide Web (WWW ’12) . Association for
Computing Machinery, New York, NY, USA, 589–598. https://doi.org/10.1145/2187836.2187916
[119] Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer,
and Veselin Stoyanov. 2019. Roberta: A Robustly Optimized Bert Pretraining Approach. arXiv preprint arXiv:1907.11692
(2019). arXiv:1907.11692
[120] Daniel Lopresti. 2009. Optical Character Recognition Errors and Their Effects on Natural Language Processing.
International Journal on Document Analysis and Recognition (IJDAR) 12, 3 (Sept. 2009), 141–151. https://doi.org/10.
1007/s10032-009-0094-8
[121] Xuezhe Ma and Eduard Hovy. 2016. End-to-End Sequence Labeling via Bi-Directional LSTM-CNNs-CRF. In Proc.
of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) . ACL, Berlin,
Germany, 1064–1074. https://doi.org/10.18653/v1/P16-1101
[122] Khai Mai, Thai-Hoang Pham, Minh Trung Nguyen, Tuan Duc Nguyen, Danushka Bollegala, Ryohei Sasano, and
Satoshi Sekine. 2018. An Empirical Study on Fine-Grained Named Entity Recognition. In Proc. of the 27th International
Conference on Computational Linguistics . ACL, Santa Fe, New Mexico, USA, 711–722. https://www.aclweb.org/
anthology/C18-1060
[123] Katja Markert and Malvina Nissim. 2009. Data and Models for Metonymy Resolution. Language Resources and
Evaluation 43, 2 (2009), 123–138.
[124] Louis Martin, Benjamin Muller, Pedro Javier Ortiz Suárez, Yoann Dupont, Laurent Romary, Éric de la Clergerie, Djamé
Seddah, and Benoît Sagot. 2020. CamemBERT: A Tasty French Language Model. In Proc. of the 58th Annual Meeting of
the Association for Computational Linguistics . ACL, Online, 7203–7219. https://www.aclweb.org/anthology/2020.acl-
main.645
[125] Sina Menzel, Josefine Zinck, Hannes Schnaitter, and Vivien Petras. 2021. Guidelines for Full Text Annotations in the
SoNAR (IDH) Corpus . Technical Report. Zenodo. https://doi.org/10.5281/zenodo.5115933
[126] Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. Efficient Estimation of Word Representations in
Vector Space. In Workshop Track Proc. Scottsdale, AZ, USA, 1–12. arXiv:1301.3781 https://arxiv.org/abs/1301.3781
[127] Diego Mollá, Menno van Zaanen, and Daniel Smith. 2006. Named Entity Recognition for Question Answering. In
Proc. of the Australasian Language Technology Workshop 2006 . ACL, Sydney, Australia, 51–58. https://www.aclweb.
org/anthology/U06-1009
[128] Ludovic Moncla, Mauro Gaio, Thierry Joliveau, and Yves-François Le Lay. 2017. Automated Geoparsing of Paris
Street Names in 19th Century Novels. In Proc. of the 1st ACM SIGSPATIAL Workshop on Geospatial Humanities . ACM,
Redondo Beach CA USA, 1–8. https://doi.org/10.1145/3149858.3149859
[129] Guenter Muehlberger et al .2019. Transforming Scholarship in the Archives through Handwritten Text Recognition:
Transkribus as a Case Study. Journal of Documentation 75, 5 (Jan. 2019), 954–976. https://doi.org/10.1108/JD-07-
2018-0114
[130] David Nadeau and Satoshi Sekine. 2007. A Survey of Named Entity Recognition and Classification. Lingvisticae
Investigationes 30, 1 (2007), 3–26. https://doi.org/10.1075/li.30.1.03nad
[131] Zara Nasar, Syed Waqar Jaffry, and Muhammad Kamran Malik. 2021. Named Entity Recognition and Relation
Extraction: State-of-the-Art. Comput. Surveys 54, 1 (Feb. 2021), 20:1–20:39. https://doi.org/10.1145/3445965
[132] Clemens Neudecker. 2016. An Open Corpus for Named Entity Recognition in Historic Newspapers. In Proc. of
the Tenth International Conference on Language Resources and Evaluation (LREC 2016) . ELRA, Portorož, Slovenia,Named Entity Recognition and Classification on Historical Documents: A Survey 47
4348–4352.
[133] Clemens Neudecker and Apostolos Antonacopoulos. 2016. Making Europe’s Historical Newspapers Searchable. In
2016 12th IAPR Workshop on Document Analysis Systems (DAS) . IEEE, Santorini, Greece, 405–410. https://doi.org/10.
1109/DAS.2016.83
[134] Clemens Neudecker, Lotte Wilms, Wille Jaan Faber, and Theo van Veen. 2014. Large-Scale Refinement of Digital
Historic Newspapers with Named Entity Recognition. In Proc IFLA Newspapers/GENLOC Pre-Conference Satellite
Meeting . IFLA, Geneva, Switzerland, 1–15.
[135] Malvina Nissim, Colin Matheson, and James Reid. 2004. Recognising Geographical Entities in Scottish Historical
Documents. In Proc. of the Workshop on Geographic Information Retrieval at SIGIR 2004 , Vol. 38. ACM, Sheffield, UK,
1–3. http://www.geo.uzh.ch/~rsp/gir/program.html
[136] Damien Nouvel, Maud Ehrmann, and Sophie Rosset. 2015. Les Entités Nommées Pour Le Traitement Automatique Des
Langues . ISTE Editions, London. 168 pages. http://infoscience.epfl.ch/record/221390
[137] Pedro Javier Ortiz Suárez, Yoann Dupont, Gaël Lejeune, and Tian Tian. 2020. SinNer@CLEF-HIPE2020: Sinful
Adaptation of SotA Models for Named Entity Recognition in Historical French and German Newspapers. In Working
Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda Cappellato, Carsten Eickhoff, Nicola Ferro, and
Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 1–12. http://ceur-ws.org/Vol-2696/paper_203.pdf
[138] Thomas L. Packer, Joshua F. Lutes, Aaron P. Stewart, David W. Embley, Eric K. Ringger, Kevin D. Seppi, and Lee S.
Jensen. 2010. Extracting Person Names from Diverse and Noisy OCR Text. In Proc. of the Fourth Workshop on Analytics
for Noisy Unstructured Text Data (AND ’10) . ACM, New York, NY, USA, 19–26. https://doi.org/10.1145/1871840.1871845
[139] Thomas Padilla. 2020. Responsible Operations: Data Science, Machine Learning, and AI in Libraries . Technical Report.
OCLC Research, USA. https://doi.org/10.25333/xk7z-9g97
[140] Sinno Jialin Pan and Qiang Yang. 2010. A Survey on Transfer Learning. IEEE Transactions on Knowledge and Data
Engineering 22, 10 (Oct. 2010), 1345–1359. https://doi.org/10.1109/TKDE.2009.191
[141] Lucia Passaro and Alessandro Lenci. 2014. “Il Piave Mormorava. . . ”: Recognizing Locations and Other Named Entities
in Italian Texts on the Great War. In Proc. of the First Italian Conference on Computational Linguistics CLiC-It 2014 &
and of the Fourth International Workshop EVALITA 2014 . Pisa University Press, Pisa, 286–290.
[142] Nita Patil, Ajay S. Patil, and B. V. Pawar. 2016. Survey of Named Entity Recognition Systems with Respect to
Indian and Foreign Languages. International Journal of Computer Applications 134, 16 (Jan. 2016), 21–26. https:
//www.ijcaonline.org/archives/volume134/number16/23999-2016908197
[143] Jeffrey Pennington, Richard Socher, and Christopher Manning. 2014. GloVe: Global Vectors for Word Representation.
InProc. of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) . ACL, Doha, Qatar,
1532–1543. https://doi.org/10.3115/v1/D14-1162
[144] Matthew Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer.
2018. Deep Contextualized Word Representations. In Proc. of the 2018 Conference of the North American Chapter of the
Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) . ACL, New Orleans,
Louisiana, 2227–2237. https://doi.org/10.18653/v1/N18-1202
[145] Eva Pfanzelter, Sarah Oberbichler, Jani Marjanen, Pierre-Carl Langlais, and Stefan Hechl. 2021. Digital Interfaces
of Historical Newspapers: Opportunities, Restrictions and Recommendations. Journal of Data Mining & Digital
Humanities HistoInformatics (2021), 1–26. https://jdmdh.episciences.org/7069
[146] Emanuele Pianta, Christian Girardi, and Roberto Zanoli. 2008. The TextPro Tool Suite. In Proc. of the International
Conference on Language Resources and Evaluation, LREC 2008, 26 May - 1 June 2008, Marrakech, Morocco . ELRA,
Marrakech, Morocco, 2603–2607. http://www.lrec-conf.org/proceedings/lrec2008/summaries/645.html
[147] Michael Piotrowski. 2012. Natural Language Processing for Historical Texts . Synthesis Lectures on Human Language
Technologies, Vol. 17. Morgan & Claypool, USA. https://doi.org/10.2200/S00436ED1V01Y201207HLT017
[148] Jakub Piskorski and Maud Ehrmann. 2013. On Named Entity Recognition in Targeted Twitter Streams in Polish.. In
Proc. of the 4th Biennial International Workshop on Balto-Slavic Natural Language Processing . ACL, Sofia, Bulgaria,
84–93. https://www.aclweb.org/anthology/W13-2413
[149] Barbara Plank. 2016. What to Do about Non-Standard (or Non-Canonical) Language in NLP. In Proc. of the 13th
Conference on Natural Language Processing (KONVENS 2016)) . Bochumer Linguistische Arbeitsberichte, Bochum,
13–20.
[150] MªLuisa Díez Platas, Salvador Ros Muñoz, Elena González-Blanco, Pablo Ruiz Fabo, and Elena Álvarez Mellado.
2020. Medieval Spanish (12th–15th Centuries) Named Entity Recognition and Attribute Annotation System Based
on Contextual Information. Journal of the Association for Information Science and Technology n/a, n/a (2020), 1–15.
https://doi.org/10.1002/asi.24399
[151] Sameer Pradhan, Alessandro Moschitti, Nianwen Xue, Olga Uryupina, and Yuchen Zhang. 2012. CoNLL-2012 Shared
Task: Modeling Multilingual Unrestricted Coreference in OntoNotes. In Joint Conference on EMNLP and CoNLL -
Shared Task (CoNLL ’12) . ACL, USA, 1–40.48 Ehrmann et al.
[152] Vera Provatorova, Svitlana Vakulenko, Evangelos Kanoulas, Koen Dercksen, and Johannes M van Hulst. 2020. Named
Entity Recognition and Linking on Historical Newspapers: UvA.ILPS & REL at CLEF HIPE 2020. In Working Notes
of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda Cappellato, Carsten Eickhoff, Nicola Ferro, and
Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 8. http://ceur-ws.org/Vol-2696/paper_209.pdf
[153] Alec Radford, Karthik Narasimhan, Tim Salimans, and Ilya Sutskever. 2018. Improving Language Understanding by
Generative Pre-Training. (2018), 1–12.
[154] Alan Ramponi and Barbara Plank. 2020. Neural Unsupervised Domain Adaptation in NLP—A Survey. In Proc. of the
28th International Conference on Computational Linguistics . International Committee on Computational Linguistics,
Barcelona, Spain (Online), 6838–6855. https://doi.org/10.18653/v1/2020.coling-main.603
[155] Lance A. Ramshaw and Mitch Marcus. 1999. Text Chunking Using Transformation-Based Learning. In Natural
Language Processing Using Very Large Corpora , Susan Armstrong, Kenneth Church, Pierre Isabelle, Sandra Manzi,
Evelyne Tzoukermann, and David Yarowsky (Eds.). Springer Netherlands, Dordrecht, 157–176. https://doi.org/10.
1007/978-94-017-2390-9_10
[156] Delip Rao, Paul McNamee, and Mark Dredze. 2013. Entity Linking: Finding Extracted Entities in a Knowledge Base.
InMulti-Source, Multilingual Information Extraction and Summarization , Thierry Poibeau, Horacio Saggion, Jakub
Piskorski, and Roman Yangarber (Eds.). Springer, Berlin, Heidelberg, 93–115. https://doi.org/10.1007/978-3-642-
28569-1_5
[157] Lev Ratinov and Dan Roth. 2009. Design Challenges and Misconceptions in Named Entity Recognition. In Proc. of the
Thirteenth Conference on Computational Natural Language Learning (CoNLL-2009) . ACL, Boulder, Colorado, 147–155.
https://www.aclweb.org/anthology/W09-1119
[158] Martin Riedl and Sebastian Padó. 2018. A Named Entity Recognition Shootout for German. In Proc. of the 56th Annual
Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) . ACL, Melbourne, Australia, 120–125.
https://doi.org/10.18653/v1/P18-2020
[159] Alan Ritter, Sam Clark, Oren Etzioni, et al .2011. Named Entity Recognition in Tweets: An Experimental Study. In Proc.
of the Conference on Empirical Methods in Natural Language Processing . ACL, Edinburgh, Scotland, UK., 1524–1534.
[160] Dominique Ritze, Cäcilia Zirn, Colin Greenstreet, Kai Eckert, and Simone Paolo Ponzetto. 2014. Named Entities
in Court: The MarineLives Corpus. In Language Resources and Technologies for Processing and Linking Historical
Documents and Archives - Deploying Linked Open Data in Cultural Heritage Workshop : Associated with the LREC
2014 Conference, 26 - 30 May 2014, Reykjavik , Kristin Bjarnadóttir (Ed.). LREC, Reykjavik, 26–30. http://www.lrec-
conf.org/proceedings/lrec2014/workshops/LREC2014Workshop-LRT4HDA%20Proc..pdf
[161] Giuseppe Rizzo and Raphaël Troncy. 2012. NERD: A Framework for Unifying Named Entity Recognition and
Disambiguation Extraction Tools. In Proc. of the Demonstrations at the 13th Conference of the European Chapter of the
Association for Computational Linguistics . ACL, Avignon, France, 73–76. https://www.aclweb.org/anthology/E12-2015
[162] Danny Rodrigues Alves, Giovanni Colavizza, and Frédéric Kaplan. 2018. Deep Reference Mining From Scholarly
Literature in the Arts and Humanities. Frontiers in Research Metrics and Analytics 3 (2018), 1–13. https://doi.org/10.
3389/frma.2018.00021
[163] Kepa Joseba Rodriquez, Mike Bryant, Tobias Blanke, and Magdalena Luszczynska. 2012. Comparison of Named
Entity Recognition Tools for Raw OCR Text. In 11th Conference on Natural Language Processing, KONVENS 2012,
Empirical Methods in Natural Language Processing, Vienna, Austria, September 19-21, 2012 (Scientific Series of the ÖGAI,
Vol. 5) , Jeremy Jancsary (Ed.). ÖGAI, Wien, Österreich, 410–414. http://www.oegai.at/konvens2012/proceedings/60_
rodriquez12w/
[164] Matteo Romanello, Maud Ehrmann, Simon Clematide, and Daniele Guido. 2020. The Impresso System Architecture
in a Nutshell. https://pro.europeana.eu/page/issue-16-newspapers#the-impresso-system-architecture-in-a-nutshell
[165] Sophie Rosset, Grouin, Cyril, Fort, Karen, Galibert, Olivier, Kahn, Juliette, and Zweigenbaum, Pierre. 2012. Structured
Named Entities in Two Distinct Press Corpora: Contemporary Broadcast News and Old Newspapers. In 6th Linguistics
Annotation Workshop (The LAW VI) . ACL, Jeju, South Korea, 40–48.
[166] Sophie Rosset, Grouin, Cyril, and Zweigenbaum, Pierre. 2011. Entités Nommées Structurées : Guide d’annotation
Quaero . Technical Report 2011-04. LIMSI-CNRS, Orsay, France.
[167] Sebastian Ruder. 2018. NLP’s ImageNet Moment Has Arrived. The Gradient (2018). https://thegradient.pub/nlp-
imagenet/
[168] Sebastian Ruder, Matthew E. Peters, Swabha Swayamdipta, and Thomas Wolf. 2019. Transfer Learning in Natural
Language Processing. In Proc. of the 2019 Conference of the North American Chapter of the Association for Computational
Linguistics: Tutorials . ACL, Minneapolis, Minnesota, 15–18. https://doi.org/10.18653/v1/N19-5004
[169] Teemu Ruokolainen and Kimmo Kettunen. 2018. À La Recherche Du Nom Perdu–Searching for Named Entities with
Stanford NER in a Finnish Historical Newspaper and Journal Collection. In 13th IAPR International Workshop on
Document Analysis Systems . IEEE Computer Society, Vienna, Austria, 1–2.Named Entity Recognition and Classification on Historical Documents: A Survey 49
[170] M. Schuster and K. K. Paliwal. 1997. Bidirectional Recurrent Neural Networks. IEEE Transactions on Signal Processing
45, 11 (Nov. 1997), 2673–2681. https://doi.org/10.1109/78.650093
[171] Stefan Schweter. 2020. Europeana BERT and ELECTRA Models. Zenodo. https://doi.org/10.5281/zenodo.4275044
[172] Stefan Schweter and Johannes Baiter. 2019. Towards Robust Named Entity Recognition for Historic German.
InProc. of the 4th Workshop on Representation Learning for NLP (RepL4NLP-2019) . ACL, Florence, Italy, 96–103.
https://doi.org/10.18653/v1/W19-4312
[173] Stefan Schweter and Luisa März. 2020. Triple E - Effective Ensembling of Embeddings and Language Models for
NER of Historical German.. In Working Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda
Cappellato, Carsten Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece,
1–13. http://ceur-ws.org/Vol-2696/paper_173.pdf
[174] Rico Sennrich, Barry Haddow, and Alexandra Birch. 2016. Neural Machine Translation of Rare Words with Subword
Units. In Proc. of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) .
ACL, Berlin, Germany, 1715–1725. https://doi.org/10.18653/v1/P16-1162
[175] Khaled Shaalan. 2013. A Survey of Arabic Named Entity Recognition and Classification. Computational Linguistics
40, 2 (Oct. 2013), 469–510. https://doi.org/10.1162/COLI_a_00178
[176] David A. Smith. 2002. Detecting Events with Date and Place Information in Unstructured Text. In Proc. of the 2nd
ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL ’02) . Association for Computing Machinery, New York, NY,
USA, 191–196. https://doi.org/10.1145/544220.544260
[177] Caroline Sporleder. 2010. Natural Language Processing for Cultural Heritage Domains. Language and Linguistics
Compass 4, 9 (2010), 750–768. https://doi.org/10.1111/j.1749-818X.2010.00230.x
[178] Rachele Sprugnoli. 2018. Arretium or Arezzo? A Neural Approach to the Identification of Place Names in Historical
Texts. In Proc. of the Fifth Italian Conference on Computational Linguistics (CLiC-It 2018) (CEUR Workshop Proc.,
Vol. 2253) , Elena Cabrio, Alessandro Mazzei, and Fabio Tamburini (Eds.). CEUR-WS, Torino, Italy, 1–6. http://ceur-
ws.org/Vol-2253/paper26.pdf
[179] Rachele Sprugnoli and Sara Tonelli. 2019. Novel Event Detection and Classification for Historical Texts. Computational
Linguistics 45, 2 (2019), 229–265. https://doi.org/10.1162/coli\_a\_00347 arXiv:https://doi.org/10.1162/coli\_a\_00347
[180] Rachele Sprugnoli, Sara Tonelli, Giovanni Moretti, and Stefano Menini. 2016. Fifty Years of European History through
the Lens of Computational Linguistics: The De Gasperi Project. IJCoL. Italian Journal of Computational Linguistics 2,
2 (Dec. 2016), 89–99. https://doi.org/10.4000/ijcol.397
[181] Ralf Steinberger, Maud Ehrmann, Júlia Pajzs, Mohamed Ebrahim, Josef Steinberger, and Marco Turchi. 2013. Multilin-
gual Media Monitoring and Text Analysis – Challenges for Highly Inflected Languages. In Text, Speech, and Dialogue ,
Ivan Habernal and Václav Matoušek (Eds.). Springer Berlin Heidelberg, Berlin, Heidelberg, 22–33.
[182] Ralf Steinberger, Bruno Pouliquen, and Erik van der Goot. 2009. An Introduction to the Europe Media Monitor Family
of Applications. In Proc. of the SIGIR 2009 Workshop (SIGIR-CLIR’2009) , Noriko Kando Fredric Gey and Jussi Karlgren
(Eds.). Association for Computing Machinery, Boston, Massachusetts, USA, 1–8. https://arxiv.org/abs/1309.5290
[183] Lichao Sun, Kazuma Hashimoto, Wenpeng Yin, Akari Asai, Jia Li, Philip Yu, and Caiming Xiong. 2020. Adv-BERT:
BERT Is Not Robust on Misspellings! Generating Nature Adversarial Samples on BERT. arXiv:2003.04985 [cs] (Feb.
2020). arXiv:2003.04985 [cs] http://arxiv.org/abs/2003.04985
[184] Wassim Swaileh, Thierry Paquet, Sébastien Adam, and Andres Rojas Camacho. 2020. A Named Entity Extraction
System for Historical Financial Data. In Document Analysis Systems (Lecture Notes in Computer Science) , Xiang
Bai, Dimosthenis Karatzas, and Daniel Lopresti (Eds.). Springer International Publishing, Cham, 324–340. https:
//doi.org/10.1007/978-3-030-57058-3_23
[185] György Szarvas, Richárd Farkas, and András Kocsor. 2006. A Multilingual Named Entity Recognition System Using
Boosting and C4.5 Decision Tree Learning Algorithms. In Discovery Science , Ljupčo Todorovski, Nada Lavrač, and
Klaus P. Jantke (Eds.). Springer Berlin Heidelberg, Berlin, Heidelberg, 267–278.
[186] Bruno Taillé, Vincent Guigue, and Patrick Gallinari. 2020. Contextualized Embeddings in Named-Entity Recognition:
An Empirical Study on Generalization. In Advances in Information Retrieval (Lecture Notes in Computer Science) ,
Joemon M. Jose, Emine Yilmaz, João Magalhães, Pablo Castells, Nicola Ferro, Mário J. Silva, and Flávio Martins (Eds.).
Springer International Publishing, Cham, 383–391. https://doi.org/10.1007/978-3-030-45442-5_48
[187] Yi Tay, Mostafa Dehghani, Dara Bahri, and Donald Metzler. 2020. Efficient Transformers: A Survey. arXiv:2009.06732
[cs](Sept. 2020), 1–28. arXiv:2009.06732 [cs] http://arxiv.org/abs/2009.06732
[188] Melissa Terras. 2011. The Rise of Digitization. In Digitisation Perspectives , Ruth Rikowski (Ed.). SensePublishers,
Rotterdam, 3–20. https://doi.org/10.1007/978-94-6091-299-3\_1
[189] Paul Thompson, Riza Theresa Batista-Navarro, Georgios Kontonatsios, Jacob Carter, Elizabeth Toon, John McNaught,
Carsten Timmermann, Michael Worboys, and Sophia Ananiadou. 2016. Text Mining the History of Medicine. PLOS
ONE 11, 1 (Jan. 2016), e0144717. https://doi.org/10.1371/journal.pone.014471750 Ehrmann et al.
[190] Erik F. Tjong Kim Sang and Fien De Meulder. 2003. Introduction to the CoNLL-2003 Shared Task: Language-
Independent Named Entity Recognition. In Proc. of the Seventh Conference on Natural Language Learning at HLT-
NAACL 2003 . Association for Computational Linguistics, ACL, Edmonton, Canada, 142–147. https://www.aclweb.
org/anthology/W03-0419
[191] Konstantin Todorov and Giovanni Colavizza. 2020. Transfer Learning for Named Entity Recognition in Historical
Corpora. In Working Notes of CLEF 2020 - Conference and Labs of the Evaluation Forum , Linda Cappellato, Carsten
Eickhoff, Nicola Ferro, and Aurélie Névéol (Eds.), Vol. 2696. CEUR-WS, Thessaloniki, Greece, 1–12. http://ceur-
ws.org/Vol-2696/paper_168.pdf
[192] Sara Tonelli, Rachele Sprugnoli, and Giovanni Moretti. 2019. Prendo La Parola in Questo Consesso Mondiale: A
Multi-Genre 20th Century Corpus in the Political Domain. In Proc. of the Sixth Italian Conference on Computational
Linguistics, Bari, Italy, November 13-15, 2019 (CEUR Workshop Proc., Vol. 2481) , Raffaella Bernardi, Roberto Navigli,
and Giovanni Semeraro (Eds.). CEUR-WS.org, Bari, Italy, 1–8. http://ceur-ws.org/Vol-2481/paper71.pdf
[193] Joseph Turian, Lev-Arie Ratinov, and Yoshua Bengio. 2010. Word Representations: A Simple and General Method for
Semi-Supervised Learning. In Proc. of the 48th Annual Meeting of the Association for Computational Linguistics . ACL,
Uppsala, Sweden, 384–394. https://www.aclweb.org/anthology/P10-1040
[194] Matje van de Camp and Antal van den Bosch. 2011. A Link to the Past: Constructing Historical Social Networks. In
Proc. of the 2nd Workshop on Computational Approaches to Subjectivity and Sentiment Analysis (WASSA 2.011) . ACL,
Portland, Oregon, 61–69. https://www.aclweb.org/anthology/W11-1708
[195] Daniel van Strien, Kaspar Beelen, Mariona Ardanuy, Kasra Hosseini, Barbara McGillivray, and Giovanni Colavizza.
2020. Assessing the Impact of OCR Quality on Downstream NLP Tasks. In Proc. of the 12th International Conference
on Agents and Artificial Intelligence . SCITEPRESS - Science and Technology Publications, Valletta, Malta, 484–496.
https://doi.org/10.5220/0009169004840496
[196] Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia
Polosukhin. 2017. Attention Is All You Need. In Advances in Neural Information Processing Systems , I. Guyon, U. V.
Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett (Eds.), Vol. 30. Curran Associates, Inc., Long
Beach, California, US, 5998–6008. https://proceedings.neurips.cc/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-
Paper.pdf
[197] Marc Vilain, Jennifer Su, and Suzi Lubar. 2007. Entity Extraction Is a Boring Solved Problem: Or Is It?. In Human
Language Technologies 2007: The Conference of the North American Chapter of the Association for Computational
Linguistics; Companion Volume, Short Papers (NAACL-Short ’07) . ACL, USA, 181–184.
[198] Miguel Won, Patricia Murrieta-Flores, and Bruno Martins. 2018. Ensemble Named Entity Recognition (NER): Evaluat-
ing NER Tools in the Identification of Place Names in Historical Corpora. Frontiers in Digital Humanities 5 (2018), 2.
https://doi.org/10.3389/fdigh.2018.00002
[199] Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V. Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun,
Yuan Cao, Qin Gao, Klaus Macherey, Jeff Klingner, Apurva Shah, Melvin Johnson, Xiaobing Liu, Łukasz Kaiser,
Stephan Gouws, Yoshikiyo Kato, Taku Kudo, Hideto Kazawa, Keith Stevens, George Kurian, Nishant Patil, Wei Wang,
Cliff Young, Jason Smith, Jason Riesa, Alex Rudnick, Oriol Vinyals, Greg Corrado, Macduff Hughes, and Jeffrey Dean.
2016. Google’s Neural Machine Translation System: Bridging the Gap between Human and Machine Translation.
arXiv:1609.08144 [cs] (Oct. 2016). arXiv:1609.08144 [cs] http://arxiv.org/abs/1609.08144
[200] Vikas Yadav and Steven Bethard. 2018. A Survey on Recent Advances in Named Entity Recognition from Deep
Learning Models. In Proc. of the 27th International Conference on Computational Linguistics . ACL, Santa Fe, New
Mexico, USA, 2145–2158. https://www.aclweb.org/anthology/C18-1182
[201] Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, and Yuji Matsumoto. 2020. LUKE: Deep Contextualized
Entity Representations with Entity-Aware Self-Attention. In Proc. of the 2020 Conference on Empirical Methods in Natural
Language Processing (EMNLP) . ACL, Online, 6442–6454. https://www.aclweb.org/anthology/2020.emnlp-main.523
[202] Jie Yang, Shuailong Liang, and Yue Zhang. 2018. Design Challenges and Misconceptions in Neural Sequence Labeling.
InProc. of the 27th International Conference on Computational Linguistics . ACL, Santa Fe, New Mexico, USA, 3879–3889.
https://aclanthology.org/C18-1327
[203] Peng Yu and Xin Wang. 2020. BERT-Based Named Entity Recognition in Chinese Twenty-Four Histories. In Web
Information Systems and Applications (Lecture Notes in Computer Science) , Guojun Wang, Xuemin Lin, James Hendler,
Wei Song, Zhuoming Xu, and Genggeng Liu (Eds.). Springer International Publishing, Cham, 289–301. https:
//doi.org/10.1007/978-3-030-60029-7_27