shibing624 commited on
Commit
3ecf529
1 Parent(s): c273d21

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +128 -1
README.md CHANGED
@@ -1,3 +1,130 @@
1
  ---
2
- license: cc-by-4.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ annotations_creators:
3
+ - shibing624
4
+ language_creators:
5
+ - shibing624
6
+ language:
7
+ - zh
8
+ license:
9
+ - cc-by-4.0
10
+ multilinguality:
11
+ - zh
12
+ size_categories:
13
+ - 100K<n<20M
14
+ source_datasets:
15
+ - https://www.biendata.xyz/competition/sohu_2021/data/
16
+ task_categories:
17
+ - text-classification
18
+ - Sentence-Similarity
19
+ task_ids:
20
+ - natural-language-inference
21
+ - semantic-similarity-scoring
22
+ - text-scoring
23
+ paperswithcode_id: sts
24
+ pretty_name: Sentence Text Similarity SOHU2021
25
  ---
26
+ # Dataset Card for NLI_zh
27
+
28
+ ## Dataset Description
29
+ - **Repository:** [Chinese NLI dataset](https://github.com/shibing624/text2vec)
30
+ - **Leaderboard:** [NLI_zh leaderboard](https://github.com/shibing624/text2vec) (located on the homepage)
31
+ - **Size of downloaded dataset files:** 218 MB
32
+ - **Total amount of disk used:** 218 MB
33
+ ### Dataset Summary
34
+
35
+ 2021搜狐校园文本匹配算法大赛数据集
36
+
37
+ - 数据源:https://www.biendata.xyz/competition/sohu_2021/data/
38
+
39
+ 分为 A 和 B 两个文件,A 和 B 文件匹配标准不一样。其中 A 和 B 文件又分为“短短文本匹配”、“短长文本匹配”和“长长文本匹配”。
40
+ A 文件匹配标准较为宽泛,两段文字是同一个话题便视为匹配,B 文件匹配标准较为严格,两段文字须是同一个事件才视为匹配。
41
+
42
+
43
+ 数据类型:
44
+
45
+
46
+ | type | 数据类型 |
47
+ | --- | ------------|
48
+ | dda | 短短匹配 A 类 |
49
+ | ddb | 短短匹配 B 类 |
50
+ | dca | 短长匹配 A 类 |
51
+ | dcb | 短长匹配 B 类 |
52
+ | cca | 长长匹配 A 类 |
53
+ | ccb | 长长匹配 B 类 |
54
+
55
+ ### Supported Tasks and Leaderboards
56
+
57
+ Supported Tasks: 支持中文文本匹配任务,文本相似度计算等相关任务。
58
+
59
+ 中文匹配任务的结果目前在顶会paper上出现较少,我罗列一个我自己训练的结果:
60
+
61
+ **Leaderboard:** [NLI_zh leaderboard](https://github.com/shibing624/text2vec)
62
+
63
+ ### Languages
64
+
65
+ 数据集均是简体中文文本。
66
+
67
+ ## Dataset Structure
68
+ ### Data Instances
69
+ An example of 'train' looks as follows.
70
+ ```python
71
+ # A 类 短短 样本示例
72
+ {
73
+ "sentence1": "小艺的故事让爱回家2021年2月16日大年初五19:30带上你最亲爱的人与团团君相约《小艺的故事》直播间!",
74
+ "sentence2": "香港代购了不起啊,宋点卷竟然在直播间“炫富”起来",
75
+ "label": 0
76
+ }
77
+
78
+ # B 类 短短 样本示例
79
+ {
80
+ "sentence1": "让很多网友好奇的是,张柏芝在一小时后也在社交平台发文:“给大家拜年啦。”还有网友猜测:谢霆锋的经纪人发文,张柏芝也发文,并且配图,似乎都在证实,谢霆锋依旧和王菲在一起,而张柏芝也有了新的恋人,并且生了孩子,两人也找到了各自的归宿,有了自己的幸福生活,让传言不攻自破。",
81
+ "sentence2": "陈晓东谈旧爱张柏芝,一个口误暴露她的秘密,难怪谢霆锋会离开她",
82
+ "label": 0
83
+ }
84
+ ```
85
+
86
+ label: 0表示不匹配,1表示匹配。
87
+
88
+ ### Data Fields
89
+ The data fields are the same among all splits.
90
+
91
+ - `sentence1`: a `string` feature.
92
+ - `sentence2`: a `string` feature.
93
+ - `label`: a classification label, with possible values including `similarity` (1), `dissimilarity` (0).
94
+
95
+ ### Data Splits
96
+
97
+
98
+ ```shell
99
+ > wc -l *.jsonl
100
+ 11690 cca.jsonl
101
+ 11690 ccb.jsonl
102
+ 11592 dca.jsonl
103
+ 11593 dcb.jsonl
104
+ 11512 dda.jsonl
105
+ 11501 ddb.jsonl
106
+ 69578 total
107
+ ```
108
+
109
+ ### Curation Rationale
110
+ 作为中文NLI(natural langauge inference)数据集,这里把这个数据集上传到huggingface的datasets,方便大家使用。
111
+
112
+ #### Who are the source language producers?
113
+ 数据集的版权归原作者所有,使用各数据集时请尊重原数据集的版权。
114
+
115
+ #### Who are the annotators?
116
+ 原作者。
117
+
118
+ ### Social Impact of Dataset
119
+ This dataset was developed as a benchmark for evaluating representational systems for text, especially including those induced by representation learning methods, in the task of predicting truth conditions in a given context.
120
+
121
+ Systems that are successful at such a task may be more successful in modeling semantic representations.
122
+
123
+ ### Licensing Information
124
+
125
+ 用于学术研究。
126
+
127
+
128
+ ### Contributions
129
+
130
+ [shibing624](https://github.com/shibing624) upload this dataset.