Datasets:

Languages:
French
ArXiv:
License:
File size: 7,868 Bytes
bdd8f0d
2714efb
bdd8f0d
 
 
 
9bab50e
bdd8f0d
9bab50e
bdd8f0d
 
 
 
 
 
 
 
 
0801961
 
 
fe42f57
3a260ca
 
 
 
 
 
 
 
 
 
 
3a62c04
 
 
3a260ca
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3a62c04
 
 
3a260ca
 
 
 
 
bdd8f0d
 
 
 
 
 
 
fe42f57
bdd8f0d
 
 
fe42f57
 
bdd8f0d
 
 
 
 
 
 
 
 
 
 
 
 
8636010
bdd8f0d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5b6157a
 
 
 
bdd8f0d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8636010
 
 
 
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
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
---
pretty_name: OrangeSum
annotations_creators:
- found
language_creators:
- found
language:
- fr
license:
- unknown
multilinguality:
- monolingual
size_categories:
- 10K<n<100K
source_datasets:
- original
task_categories:
- summarization
task_ids:
- news-articles-headline-generation
- news-articles-summarization
paperswithcode_id: orangesum
dataset_info:
- config_name: abstract
  features:
  - name: text
    dtype: string
  - name: summary
    dtype: string
  splits:
  - name: train
    num_bytes: 53531651
    num_examples: 21401
  - name: test
    num_bytes: 3785207
    num_examples: 1500
  - name: validation
    num_bytes: 3698650
    num_examples: 1500
  download_size: 23058350
  dataset_size: 61015508
- config_name: title
  features:
  - name: text
    dtype: string
  - name: summary
    dtype: string
  splits:
  - name: train
    num_bytes: 65225136
    num_examples: 30659
  - name: test
    num_bytes: 3176690
    num_examples: 1500
  - name: validation
    num_bytes: 3276713
    num_examples: 1500
  download_size: 27321627
  dataset_size: 71678539
---

# Dataset Card for OrangeSum

## 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

- **Repository:** [OrangeSum repository](https://github.com/Tixierae/OrangeSum)
- **Paper:** [BARThez: a Skilled Pretrained French Sequence-to-Sequence Model](https://arxiv.org/abs/2010.12321)	
- **Point of Contact:** [Antoine J.-P. Tixier](Antoine.Tixier-1@colorado.edu)

### Dataset Summary

The OrangeSum dataset was inspired by the XSum dataset. It was created by scraping the "Orange Actu" website: https://actu.orange.fr/. Orange S.A. is a large French multinational telecommunications corporation, with 266M customers worldwide. Scraped pages cover almost a decade from Feb 2011 to Sep 2020. They belong to five main categories: France, world, politics, automotive, and society. The society category is itself divided into 8 subcategories: health, environment, people, culture, media, high-tech, unsual ("insolite" in French), and miscellaneous.

Each article featured a single-sentence title as well as a very brief abstract, both professionally written by the author of the article. These two fields were extracted from each page, thus creating two summarization tasks: OrangeSum Title and OrangeSum Abstract.

### Supported Tasks and Leaderboards

**Tasks:** OrangeSum Title and OrangeSum Abstract.

To this day, there is no Leaderboard for this dataset.

### Languages

The text in the dataset is in French.

## Dataset Structure

### Data Instances

A data instance consists of a news article  and a summary. The summary can be a short abstract or a title depending on the configuration.

Example:

**Document:** Le temps sera pluvieux sur huit départements de la France ces prochaines heures : outre les trois départements bretons placés en vigilance orange jeudi matin, cinq autres départements du sud du Massif Central ont été à leur tour placés en alerte orange pluie et inondation. Il s'agit de l'Aveyron, du Cantal, du Gard, de la Lozère, et de la Haute-Loire. Sur l'ensemble de l'épisode, les cumuls de pluies attendus en Bretagne sont compris entre 40 et 60 mm en 24 heures et peuvent atteindre localement les 70 mm en 24 heures.Par la suite, la dégradation qui va se mettre en place cette nuit sur le Languedoc et le sud du Massif Central va donner sur l'Aveyron une première salve intense de pluie. Des cumuls entre 70 et 100 mm voir 120 mm localement sont attendus sur une durée de 24 heures. Sur le relief des Cévennes on attend de 150 à 200 mm, voire 250 mm très ponctuellement sur l'ouest du Gard et l'est de la Lozère. Cet épisode va s'estomper dans la soirée avec le décalage des orages vers les régions plus au nord. Un aspect orageux se mêlera à ces précipitations, avec de la grêle possible, des rafales de vent et une forte activité électrique.

**Abstract:** Outre les trois départements bretons, cinq autres départements du centre de la France ont été placés en vigilance orange pluie-inondation.

**Title:** Pluie-inondations : 8 départements en alerte orange.

### Data Fields

`text`: the document to be summarized. \
`summary`: the summary of the source document.

### Data Splits

The data is split into a training, validation and test in both configuration.

|          | train | validation | test |
|----------|------:|-----------:|-----:|
| Abstract | 21400 |       1500 | 1500 |
| Title    | 30658 |       1500 | 1500 |

## Dataset Creation

### Curation Rationale

The goal here was to create a French equivalent of the recently introduced [XSum](https://github.com/EdinburghNLP/XSum/tree/master/XSum-Dataset) dataset. Unlike the historical summarization datasets, CNN, DailyMail, and NY Times, which favor extractive strategies, XSum, as well as OrangeSum require the models to display a high degree of abstractivity to perform well. The summaries in OrangeSum are not catchy headlines, but rather capture the gist of the articles.

### Source Data

#### Initial Data Collection and Normalization

Each article features a single-sentence title as well as a very brief abstract. Extracting these two fields from each news article page, creates two summarization tasks: OrangeSum Title and OrangeSum Abstract. As a post-processing step, all empty articles and those whose summaries were shorter than 5 words were removed. For OrangeSum Abstract, the top 10% articles in terms of proportion of novel unigrams in the abstracts were removed, as it was observed that such abstracts tend to be introductions rather than real abstracts. This corresponded to a threshold of 57% novel unigrams. For both OrangeSum Title and OrangeSum Abstract, 1500 pairs for testing and 1500 for validation are set aside, and all the remaining ones are used for training.

#### Who are the source language producers?

The authors of the artiles.

### Annotations

#### Annotation process

The smmaries are professionally written by the author of the articles.

#### Who are the annotators?

The authors of the artiles.

### 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

The dataset was initially created by Antoine J.-P. Tixier.

### Licensing Information

[More Information Needed]

### Citation Information

```
@article{eddine2020barthez,
  title={BARThez: a Skilled Pretrained French Sequence-to-Sequence Model},
  author={Eddine, Moussa Kamal and Tixier, Antoine J-P and Vazirgiannis, Michalis},
  journal={arXiv preprint arXiv:2010.12321},
  year={2020}
}
```

### Contributions

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