File size: 5,401 Bytes
f25be5d
 
 
 
 
ed49445
f25be5d
ed49445
f25be5d
 
 
 
 
 
 
 
 
 
 
d0ee1ac
2f2f25c
05d606e
 
 
 
 
 
 
 
 
 
70f6170
 
05d606e
 
 
 
c542eb7
 
 
05d606e
 
 
 
 
f25be5d
 
 
 
 
 
 
d0ee1ac
f25be5d
 
 
d0ee1ac
 
f25be5d
 
 
 
 
 
 
 
 
 
 
 
 
763f1b6
f25be5d
 
 
a07178f
 
 
f25be5d
a07178f
f25be5d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
763f1b6
 
 
05d606e
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
---
annotations_creators:
- machine-generated
language_creators:
- found
language:
- tl
license:
- unknown
multilinguality:
- monolingual
size_categories:
- 100K<n<1M
source_datasets:
- original
task_categories:
- text-classification
task_ids:
- natural-language-inference
paperswithcode_id: newsph-nli
pretty_name: NewsPH NLI
dataset_info:
  features:
  - name: premise
    dtype: string
  - name: hypothesis
    dtype: string
  - name: label
    dtype:
      class_label:
        names:
          '0': '0'
          '1': '1'
  splits:
  - name: train
    num_bytes: 154510599
    num_examples: 420000
  - name: test
    num_bytes: 3283665
    num_examples: 9000
  - name: validation
    num_bytes: 33015530
    num_examples: 90000
  download_size: 76565287
  dataset_size: 190809794
---

# Dataset Card for NewsPH NLI

## Table of Contents
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Fields](#data-fields)
  - [Data Splits](#data-splits)
- [Dataset Creation](#dataset-creation)
  - [Curation Rationale](#curation-rationale)
  - [Source Data](#source-data)
  - [Annotations](#annotations)
  - [Personal and Sensitive Information](#personal-and-sensitive-information)
- [Considerations for Using the Data](#considerations-for-using-the-data)
  - [Social Impact of Dataset](#social-impact-of-dataset)
  - [Discussion of Biases](#discussion-of-biases)
  - [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
  - [Dataset Curators](#dataset-curators)
  - [Licensing Information](#licensing-information)
  - [Citation Information](#citation-information)
  - [Contributions](#contributions)

## Dataset Description

- **Homepage:** [NewsPH NLI homepage](https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks)
- **Repository:** [NewsPH NLI repository](https://github.com/jcblaisecruz02/Filipino-Text-Benchmarks)
- **Paper:** [Arxiv paper](https://arxiv.org/pdf/2010.11574.pdf)
- **Leaderboard:**
- **Point of Contact:** [Jan Christian Cruz](mailto:jan_christian_cruz@dlsu.edu.ph)

### Dataset Summary

First benchmark dataset for sentence entailment in the low-resource Filipino language. Constructed through exploting the structure of news articles. Contains 600,000 premise-hypothesis pairs, in 70-15-15 split for training, validation, and testing.


### Supported Tasks and Leaderboards

[More Information Needed]

### Languages

The dataset contains news articles in Filipino (Tagalog) scraped rom all major Philippine news sites online.

## Dataset Structure

### Data Instances
Sample data:
  {
    "premise": "Alam ba ninyo ang ginawa ni Erap na noon ay lasing na lasing na rin?",
    "hypothesis": "Ininom niya ang alak na pinagpulbusan!",
    "label": "0"
  }


### Data Fields

[More Information Needed]

### Data Splits
Contains 600,000 premise-hypothesis pairs, in 70-15-15 split for training, validation, and testing.


## Dataset Creation

### Curation Rationale

We propose the use of news articles for automatically creating benchmark datasets for NLI because of two reasons. First, news articles commonly use single-sentence paragraphing, meaning every paragraph in a news article is limited to a single sentence. Second, straight news articles follow the “inverted pyramid” structure, where every succeeding paragraph builds upon the premise of those that came before it, with the most important information on top and the least important towards the end.

### Source Data

#### Initial Data Collection and Normalization

To create the dataset, we scrape news articles from all major Philippine news sites online. We collect a total of 229,571 straight news articles, which we then lightly preprocess to remove extraneous unicode characters and correct minimal misspellings. No further preprocessing is done to preserve information in the data.

#### Who are the source language producers?

The dataset was created by Jan Christian, Blaise Cruz, Jose Kristian Resabal, James Lin, Dan John Velasco, and Charibeth Cheng from De La Salle University and the University of the Philippines

### Annotations

#### Annotation process

[More Information Needed]

#### Who are the annotators?

Jan Christian Blaise Cruz, Jose Kristian Resabal, James Lin, Dan John Velasco and Charibeth Cheng

### Personal and Sensitive Information

[More Information Needed]

## Considerations for Using the Data

### Social Impact of Dataset

[More Information Needed]

### Discussion of Biases

[More Information Needed]

### Other Known Limitations

[More Information Needed]

## Additional Information

### Dataset Curators

[Jan Christian Blaise Cruz] (mailto:jan_christian_cruz@dlsu.edu.ph)

### Licensing Information

[More Information Needed]

### Citation Information

@article{cruz2020investigating,
  title={Investigating the True Performance of Transformers in Low-Resource Languages: A Case Study in Automatic Corpus Creation},
  author={Jan Christian Blaise Cruz and Jose Kristian Resabal and James Lin and Dan John Velasco and Charibeth Cheng},
  journal={arXiv preprint arXiv:2010.11574},
  year={2020}
}

### Contributions

Thanks to [@anaerobeth](https://github.com/anaerobeth) for adding this dataset.