Update readme
Browse files
README.md
ADDED
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: cc-by-nc-sa-4.0
|
3 |
+
size_categories:
|
4 |
+
- 10K<n<100K
|
5 |
+
---
|
6 |
+
|
7 |
+
## Arxiver Dataset
|
8 |
+
Arxiver consists of 63,357 [arXiv](https://arxiv.org/) papers converted to multi-markdown (**.mmd**) format. Our dataset includes original arXiv article IDs, titles, abstracts, authors, publication dates, URLs and corresponding markdown files published between January 2023 and October 2023.
|
9 |
+
|
10 |
+
We hope our dataset will be useful for various applications such as semantic search, domain specific language modeling, question answering and summarization.
|
11 |
+
|
12 |
+
## Curation
|
13 |
+
The Arxiver dataset is created using a neural OCR - [Nougat](https://facebookresearch.github.io/nougat/). After OCR processing, we apply custom text processing steps to refine the data. This includes extracting author information, removing reference sections, and performing additional cleaning and formatting.
|
14 |
+
|
15 |
+
## Using Arxiver
|
16 |
+
You can easily download and use the arxiver dataset with Hugging Face's [datasets](https://huggingface.co/datasets) library.
|
17 |
+
```py
|
18 |
+
from datasets import load_dataset
|
19 |
+
|
20 |
+
# whole dataset takes 1.44GB
|
21 |
+
dataset = load_dataset("neuralwork/arxiver")
|
22 |
+
print(dataset)
|
23 |
+
```
|
24 |
+
|
25 |
+
Alternatively, you can stream the dataset to save disk space or to partially download the dataset:
|
26 |
+
```py
|
27 |
+
from datasets import load_dataset
|
28 |
+
|
29 |
+
dataset = load_dataset("neuralwork/arxiver", streaming=True)
|
30 |
+
print(dataset)
|
31 |
+
print(next(iter(dataset['train'])))
|
32 |
+
```
|
33 |
+
|
34 |
+
## References
|
35 |
+
The original articles are maintained by [arXiv](https://arxiv.org/) and copyrighted to the original authors, please refer to the arXiv license information [page](https://info.arxiv.org/help/license/index.html) for details. We release our dataset with a Creative Commons Attribution-Noncommercial-ShareAlike (CC BY-NC-SA 4.0) license, if you use this dataset in your research or project, please cite it as follows:
|
36 |
+
```
|
37 |
+
@misc{acar_arxiver2024,
|
38 |
+
author = {Alican Acar, Alara Dirik, Muhammet Hatipoglu},
|
39 |
+
title = {ArXiver},
|
40 |
+
year = {2024},
|
41 |
+
publisher = {Hugging Face},
|
42 |
+
howpublished = {\url{https://huggingface.co/datasets/neuralwork/arxiver}}
|
43 |
+
}
|
44 |
+
```
|