Datasets:
GEM
/

ArXiv:
License:
Tahmid commited on
Commit
dbccc03
1 Parent(s): d1b54be

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +18 -13
README.md CHANGED
@@ -251,7 +251,7 @@ Yoruba | yo | https://www.bbc.com/yoruba | 6350 | 793 | 793 | 7936 |
251
 
252
  ### Curation Rationale
253
 
254
- [More information needed](https://github.com/csebuetnlp/xl-sum)
255
 
256
  ### Source Data
257
 
@@ -259,50 +259,55 @@ Yoruba | yo | https://www.bbc.com/yoruba | 6350 | 793 | 793 | 7936 |
259
 
260
  #### Initial Data Collection and Normalization
261
 
262
- [Detailed in the paper](https://aclanthology.org/2021.findings-acl.413/)
263
 
 
 
 
 
 
264
 
265
- #### Who are the source language producers?
266
 
267
- [Detailed in the paper](https://aclanthology.org/2021.findings-acl.413/)
268
 
 
269
 
270
- ### Annotations
271
 
272
- [Detailed in the paper](https://aclanthology.org/2021.findings-acl.413/)
273
 
 
274
 
275
  #### Annotation process
276
 
277
- [Detailed in the paper](https://aclanthology.org/2021.findings-acl.413/)
278
 
279
  #### Who are the annotators?
280
 
281
- [Detailed in the paper](https://aclanthology.org/2021.findings-acl.413/)
282
 
283
  ### Personal and Sensitive Information
284
 
285
- [More information needed](https://github.com/csebuetnlp/xl-sum)
286
 
287
  ## Considerations for Using the Data
288
 
289
  ### Social Impact of Dataset
290
 
291
- [More information needed](https://github.com/csebuetnlp/xl-sum)
 
292
 
293
  ### Discussion of Biases
294
 
295
- [More information needed](https://github.com/csebuetnlp/xl-sum)
296
 
297
  ### Other Known Limitations
298
 
299
- [More information needed](https://github.com/csebuetnlp/xl-sum)
300
 
301
  ## Additional Information
302
 
303
  ### Dataset Curators
304
 
305
- [More information needed](https://github.com/csebuetnlp/xl-sum)
306
 
307
  ### Licensing Information
308
 
 
251
 
252
  ### Curation Rationale
253
 
254
+ State-of-the-art text summarization models are heavily data-driven, i.e., a large number of article-summary pairs are required to train them effectively. As a result, abstractive summarization has centered around the English language, as most large abstractive summarization datasets are available in English only. Though there have been some recent efforts for curating multilingual abstractive summarization datasets, they are limited in terms of the number of languages covered, the number of training samples, or both. To this end, we curate **XL-Sum**, a large-scale abstractive summarization dataset of 1.35 million news articles from 45 languages crawled from the British Broadcasting Corporation website.
255
 
256
  ### Source Data
257
 
 
259
 
260
  #### Initial Data Collection and Normalization
261
 
262
+ We designed a crawler to recursively crawl pages starting from the homepage by visiting different article links present in each page visited. We were able to take advantage of the fact that all BBC sites have somewhat similar structures, and were able to scrape articles from all sites. We discarded pages with no textual contents (mostly pages consisting of multimedia contents) before further processing. We designed a number of heuristics to make the extraction effective by carefully examining the HTML structures of the crawled pages:
263
 
264
+ 1. The desired summary must be present within the beginning two paragraphs of an article.
265
+ 2. The summary paragraph must have some portion of texts in bold format.
266
+ 3. The summary paragraph may contain some hyperlinks that may not be bold. The proportion of bold texts and hyperlinked texts to the total length of the paragraph in consideration must be at least 95\%.
267
+ 4. All texts except the summary and the headline must be included in the input text (including image captions).
268
+ 5. The input text must be at least twice as large as the summary.
269
 
 
270
 
 
271
 
272
+ #### Who are the source language producers?
273
 
274
+ [BBC News Editorial Team](https://www.bbc.co.uk/ws/languages)
275
 
 
276
 
277
+ ### Annotations
278
 
279
  #### Annotation process
280
 
281
+ BBC typically provides a summary of a whole article in the form of a bold paragraph containing one or two sentences at the beginning of each article. These summaries are written professionally by the authors of the articles in order to convey its main story within one small paragraph. This is in contrast to the headline which serves to draw the attention of viewers into reading the article. We used the bold texts as summary and the rest of the article as input.
282
 
283
  #### Who are the annotators?
284
 
285
+ [BBC News Editorial Team](https://www.bbc.co.uk/ws/languages)
286
 
287
  ### Personal and Sensitive Information
288
 
289
+ Meta-information like author names are discarded. However, we cannot guarantee removal of all personal information.
290
 
291
  ## Considerations for Using the Data
292
 
293
  ### Social Impact of Dataset
294
 
295
+ We believe that our efforts in this work will encourage the community to push the boundaries of abstractive text summarization beyond the English language, especially for low and mid-resource languages, bringing technological advances to communities of these languages that have been traditionally under-served.
296
+
297
 
298
  ### Discussion of Biases
299
 
300
+ Human evaluation showed most languages had a high percentage of good summaries in the upper nineties, almost none of the summaries contained any conflicting information, while about one-third on average had information that was not directly inferrable from the source article.
301
 
302
  ### Other Known Limitations
303
 
304
+ The dataset is limited to news domain only.
305
 
306
  ## Additional Information
307
 
308
  ### Dataset Curators
309
 
310
+ [Authors of this paper](https://aclanthology.org/2021.findings-acl.413)
311
 
312
  ### Licensing Information
313