taishi-i commited on
Commit
b750b28
1 Parent(s): a2358dc

update README.md

Browse files
Files changed (1) hide show
  1. README.md +139 -0
README.md CHANGED
@@ -1,3 +1,142 @@
1
  ---
2
  license: other
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: other
3
+ task_categories:
4
+ - text-classification
5
+ language:
6
+ - en
7
+ - ja
8
+ tags:
9
+ - code
10
+ size_categories:
11
+ - 1K<n<10K
12
  ---
13
+
14
+
15
+ # Dataset overview
16
+
17
+ This dataset identifies whether a GitHub repository description pertains to Japanese natural language processing (NLP).
18
+ The labels are categorized as **"Relevant (1)" and "Not Relevant (0)"**.
19
+
20
+ Problem Setting:
21
+ - Training Data: Repository descriptions from before 2022
22
+ - Test Data: Repository descriptions from 2023
23
+ - Objective: To detect repositories related to Japanese NLP
24
+
25
+ Data Collection:
26
+ - Positive Examples: Repositories listed in "[awesome-japanese-nlp-resources](https://github.com/taishi-i/awesome-japanese-nlp-resources)" as of September 9, 2023
27
+ - Negative Examples: Collected from the GitHub API and visually confirmed
28
+ - Note: The annotation process is subjective
29
+
30
+ Dataset Features:
31
+ - Subjective labeling
32
+ - Mixed English and Japanese descriptions
33
+ - Imbalanced label distribution
34
+
35
+ **These dataset features mirror real-world challenges and are ideal for evaluating models.**
36
+ Based on GitHub's terms of service, please use this dataset for research purposes only.
37
+
38
+
39
+ # How to use this dataset
40
+
41
+ How to load in Python.
42
+
43
+ ```python
44
+ from datasets import load_dataset
45
+
46
+ dataset = load_dataset("taishi-i/awesome-japanese-nlp-classification-dataset")
47
+ ```
48
+
49
+ Details of the dataset.
50
+
51
+ ```python
52
+ DatasetDict({
53
+ train: Dataset({
54
+ features: ['label', 'text', 'url', 'created_at'],
55
+ num_rows: 5496
56
+ })
57
+ validation: Dataset({
58
+ features: ['label', 'text', 'url', 'created_at'],
59
+ num_rows: 400
60
+ })
61
+ test: Dataset({
62
+ features: ['label', 'text', 'url', 'created_at'],
63
+ num_rows: 856
64
+ })
65
+ })
66
+ ```
67
+
68
+ # Baseline
69
+
70
+ Baseline trained with [bert-base-multilingual-cased](https://huggingface.co/bert-base-multilingual-cased).
71
+ The F1-score for label 1 is important for this task.
72
+
73
+ | Label | Precision | Recall | F1-Score | Support |
74
+ |--------------|-----------|--------|----------|---------|
75
+ | 0 | 0.98 | 0.99 | 0.98 | 796 |
76
+ | 1 | 0.79 | 0.70 | **0.74** | 60 |
77
+ | Accuracy | | | 0.97 | 856 |
78
+ | Macro Avg | 0.89 | 0.84 | 0.86 | 856 |
79
+ | Weighted Avg | 0.96 | 0.97 | 0.97 | 856 |
80
+
81
+
82
+
83
+ # Dataset stats
84
+
85
+ Label distribution:
86
+
87
+ | Dataset | Label 0 (%) | Label 1 (%) |
88
+ |------------|-------------|-------------|
89
+ | Train | 92.59 | 7.41 |
90
+ | Validation | 95.75 | 4.25 |
91
+ | Test | 92.99 | 7.01 |
92
+
93
+ Relevant sample:
94
+
95
+ ```python
96
+ {
97
+ "label": 1,
98
+ "text": "JGLUE: Japanese General Language Understanding Evaluation for huggingface datasets",
99
+ "url": "https://github.com/shunk031/huggingface-datasets_JGLUE",
100
+ "created_at": "2023-02-25T04:33:03Z"
101
+ }
102
+ ```
103
+
104
+ Not Relevant sample:
105
+
106
+ ```python
107
+ {
108
+ "label": 0,
109
+ "text": "Official repository of FaceLit: Neural 3D Relightable Faces (CVPR 2023)",
110
+ "url": "https://github.com/apple/ml-facelit",
111
+ "created_at": "2023-04-03T22:47:29Z"
112
+ }
113
+ ```
114
+
115
+ Number of texts, average number of characters per text, minimum number of characters, maximum number of characters:
116
+
117
+ | Dataset | Text Count | Average Length | Min Length | Max Length |
118
+ |------------|------------|----------------|------------|------------|
119
+ | Train | 5496 | 58.05 | 2.0 | 609.0 |
120
+ | Validation | 400 | 54.33 | 8.0 | 226.0 |
121
+ | Test | 856 | 58.85 | 3.0 | 341.0 |
122
+
123
+ Proportion of text languages:
124
+
125
+ | Dataset | English (%) | Japanese (%) |
126
+ |------------|-------------|--------------|
127
+ | Train | 89.34 | 10.66 |
128
+ | Validation | 82.00 | 18.00 |
129
+ | Test | 83.18 | 16.82 |
130
+
131
+ Time range:
132
+
133
+ | Dataset | Start Date | End Date |
134
+ |---------|---------------------------|---------------------------|
135
+ | Train | 2008-02-11 22:55:26+00:00 | 2022-09-30 19:45:09+00:00 |
136
+ | Validation | 2022-10-01 06:02:56+00:00 | 2022-12-31 12:12:41+00:00 |
137
+ | Test | 2023-01-01 06:15:03+00:00 | 2023-08-21 15:30:53+00:00 |
138
+
139
+
140
+ # License
141
+
142
+ We collect and publish this dataset under [GitHub Acceptable Use Policies - 7. Information Usage Restrictions](https://docs.github.com/en/site-policy/acceptable-use-policies/github-acceptable-use-policies#7-information-usage-restrictions) and [GitHub Terms of Service - H. API Terms](https://docs.github.com/en/site-policy/github-terms/github-terms-of-service#h-api-terms) for research purposes. This dataset should be used solely for research verification purposes. Adhering to GitHub's regulations is mandatory.