|
--- |
|
annotations_creators: |
|
- unknown |
|
language_creators: |
|
- unknown |
|
language: |
|
- en |
|
license: |
|
- cc-by-4.0 |
|
multilinguality: |
|
- monolingual |
|
task_categories: |
|
- text-generation |
|
size_categories: |
|
- 100K<n<1M |
|
pretty_name: KPTimes |
|
tags: |
|
- keyphrase-generation |
|
--- |
|
|
|
# KPTimes Benchmark Dataset for Keyphrase Generation |
|
|
|
## About |
|
|
|
KPTimes is a dataset for benchmarking keyphrase extraction and generation models. |
|
The dataset is composed of 290K news articles in English collected from the [New York Times](https://www.nytimes.com/) and the [Japan |
|
Times](https://www.japantimes.co.jp/). |
|
Keyphrases were annotated by editors in a semi-automated manner (that is, editors revise a set of keyphrases proposed by an algorithm and provide additional keyphrases). |
|
Details about the dataset can be found in the original paper [(Gallina et al., 2019)][gallina-2019]. |
|
|
|
Reference (indexer-assigned) keyphrases are also categorized under the PRMU (<u>P</u>resent-<u>R</u>eordered-<u>M</u>ixed-<u>U</u>nseen) scheme as proposed in [(Boudin and Gallina, 2021)][boudin-2021]. |
|
|
|
Text pre-processing (tokenization) is carried out using `spacy` (`en_core_web_sm` model) with a special rule to avoid splitting words with hyphens (e.g. graph-based is kept as one token). |
|
Stemming (Porter's stemmer implementation provided in `nltk`) is applied before reference keyphrases are matched against the source text. |
|
Details about the process can be found in `prmu.py`. <u>P</u>resent keyphrases are ordered according to their first occurrence position in the text. |
|
|
|
## Content and statistics |
|
|
|
The dataset contains the following test split: |
|
|
|
| Split | # documents | #words | # keyphrases | % Present | % Reordered | % Mixed | % Unseen | |
|
| :--------- | ----------: | -----: | -----------: | --------: | ----------: | ------: | -------: | |
|
| Train | 259,923 | 921 | 5.03 | 45.61 | 15.57 | 29.63 | 9.19 | |
|
| Validation | 10,000 | 921 | 5.02 | 45.22 | 15.78 | 29.60 | 9.41 | |
|
| Test | 20,000 | 648 | 5.03 | 60.64 | 8.90 | 18.95 | 11.51 | |
|
|
|
The following data fields are available : |
|
|
|
- **id**: unique identifier of the document. |
|
- **title**: title of the document. |
|
- **abstract**: abstract of the document. |
|
- **keyphrases**: list of reference keyphrases. |
|
- **prmu**: list of <u>P</u>resent-<u>R</u>eordered-<u>M</u>ixed-<u>U</u>nseen categories for reference keyphrases. |
|
- **date**: publishing date (YYYY/MM/DD) |
|
- **categories**: categories of the article (1 or 2 categories) |
|
|
|
## References |
|
|
|
- (Gallina et al., 2019) Ygor Gallina, Florian Boudin, and Beatrice Daille. 2019. |
|
[KPTimes: A Large-Scale Dataset for Keyphrase Generation on News Documents][gallina-2019]. |
|
In Proceedings of the 12th International Conference on Natural Language Generation, pages 130–135, Tokyo, Japan. Association for Computational Linguistics. |
|
- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021. |
|
[Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021]. |
|
In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics. |
|
|
|
[gallina-2019]: https://aclanthology.org/W19-8617/ |
|
[boudin-2021]: https://aclanthology.org/2021.naacl-main.330/ |
