hungnm commited on
Commit
08b1866
·
verified ·
1 Parent(s): ebe473f

Update README

Browse files
Files changed (1) hide show
  1. README.md +114 -0
README.md CHANGED
@@ -32,4 +32,118 @@ configs:
32
  path: data/validation-*
33
  - split: test
34
  path: data/test-*
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
35
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
  path: data/validation-*
33
  - split: test
34
  path: data/test-*
35
+ license: apache-2.0
36
+ task_categories:
37
+ - text-classification
38
+ language:
39
+ - ar
40
+ - de
41
+ - en
42
+ - es
43
+ - fr
44
+ - hi
45
+ - id
46
+ - it
47
+ - ko
48
+ - ms
49
+ - pt
50
+ - ru
51
+ - tr
52
+ - vi
53
+ - zh
54
+ - ja
55
+ tags:
56
+ - sentiment
57
+ - multilingual
58
+ - emotion
59
+ - review
60
+ - classification
61
+ pretty_name: text
62
+ size_categories:
63
+ - 1M<n<10M
64
  ---
65
+ ## Overview
66
+ **MultilingualSentiment** is a sentiment classification dataset that encompasses three sentiment labels: **Positive**, **Neutral**, **Negative**
67
+
68
+ The dataset spans multiple languages and covers a wide range of domains, making it ideal for multilingual sentiment analysis tasks.
69
+
70
+
71
+ ## Dataset Information
72
+ The dataset was meticulously collected and aggregated from various sources, including Hugging Face and Kaggle. These sources provide diverse languages and domains to ensure a comprehensive and balanced dataset.
73
+
74
+ - **Total records**: 3,934,349
75
+ - The dataset is divided into three subsets: train, validation, and test, with a ratio of 8:1:1:
76
+ + Train: 3,147,478
77
+ + Validation: 393,435
78
+ + Test: 393,436
79
+
80
+ ### Number of Records per Language
81
+ | Language | Count |
82
+ |---------------|---------|
83
+ | Arabic (ar) | 208,375 |
84
+ | German (de) | 212,853 |
85
+ | English (en) | 1,519,860 |
86
+ | Spanish (es) | 222,911 |
87
+ | French (fr) | 262,645 |
88
+ | Hindi (hi) | 9,423 |
89
+ | Indonesian (id) | 12,536 |
90
+ | Italian (it) | 3,020 |
91
+ | Japanese (ja) | 335,656 |
92
+ | Korean (ko) | 259,998 |
93
+ | Malay (ms) | 6,661 |
94
+ | Multilingual | 9,391 |
95
+ | Portuguese (pt) | 49,188 |
96
+ | Russian (ru) | 205,186 |
97
+ | Turkish (tr) | 44,743 |
98
+ | Vietnamese (vi) | 127,068 |
99
+ | Chinese (zh) | 444,835 |
100
+
101
+ ### Number of Records per Label
102
+ | Label | Count |
103
+ |-----------|----------|
104
+ | Negative | 1,436,539 |
105
+ | Neutral | 1,041,512 |
106
+ | Positive | 1,456,298 |
107
+
108
+ ## Applications
109
+ This dataset is well-suited for training and evaluating models in multilingual sentiment analysis, natural language processing (NLP), and domain-specific sentiment classification tasks.
110
+
111
+ ## Loading dataset
112
+ ```python
113
+ from datasets import load_dataset
114
+
115
+ # Load the MultilingualSentiment dataset
116
+ dataset = load_dataset("clapAI/MultiLingualSentiment")
117
+
118
+ print(dataset)
119
+
120
+ ```
121
+
122
+ ```
123
+ DatasetDict({
124
+ train: Dataset({
125
+ features: ['text', 'label', 'source', 'domain', 'language'],
126
+ num_rows: 3147478
127
+ })
128
+ validation: Dataset({
129
+ features: ['text', 'label', 'source', 'domain', 'language'],
130
+ num_rows: 393435
131
+ })
132
+ test: Dataset({
133
+ features: ['text', 'label', 'source', 'domain', 'language'],
134
+ num_rows: 393436
135
+ })
136
+ })
137
+ ```
138
+
139
+ ## Citation
140
+
141
+ ```bibtex
142
+ @dataset{clapAI2024multilingualsentiment,
143
+ title = {MultilingualSentiment: A Multilingual Sentiment Classification Dataset},
144
+ author = {clapAI},
145
+ year = {2024},
146
+ url = {https://huggingface.co/datasets/clapAI/MultiLingualSentiment},
147
+ description = {A multilingual dataset for sentiment analysis with labels: positive, neutral, negative, covering diverse languages and domains.},
148
+ }
149
+ ```