Biographical Relation Extraction Dataset
Welcome to the repository of datasets tailored for biographical relation extraction, crafted utilizing Guided Distant Supervision (GDS). Explore datasets available in both English and German, which facilitate extensive research in relation extraction from biographical data. Below you can find an overview of the datasets currently available, as well as the relations that are in each set. Please note there are different sets for each language, which denote how they were compiled. In short, normal followed GDS, coref added coreference resolution and skip skipped certain parts of the text. For a more extensive explanation how this worked, please refer to [1].
Available Datasets
English Dataset
Overview
Detailed insights into the English dataset can be found in [1].
Data Summary
| Relation | Normal Set | Coref Set | Skip Set |
|---|---|---|---|
| Birthdate | 51,524 | 47,977 | 45,211 |
| Birthplace | 50,226 | 46,551 | 17,537 |
| Deathdate | 17,197 | 14,500 | 5,925 |
| Deathplace | 18,944 | 20,430 | 10,790 |
| Occupation | 18,114 | 18,111 | 8,716 |
| Parent | 6,352 | 10,291 | 5,596 |
| Educated | 5,639 | 9,415 | 3,858 |
| Child | 2,209 | 4,053 | 2,123 |
| Sibling | 2,083 | 3,601 | 1,997 |
| Other | 173,969 | 175,916 | 103,248 |
| Total | 346,257 | 350,845 | 205,001 |
German Dataset
Overview
A paper discussing the German dataset is forthcoming [2].
Data Summary
| Relation | Normal Set | Skip Set |
|---|---|---|
| Birthdate | 8,777 | 770 |
| Birthplace | 12,833 | 5,816 |
| Child | 718 | 701 |
| Deathdate | 922 | 454 |
| Deathplace | 4,059 | 3,263 |
| Educated | 610 | 607 |
| Occupation | 10,861 | 4,836 |
| Other | 39,782 | 20,469 |
| Parent | 3,704 | 3,565 |
| Sibling | 917 | 890 |
| Total | 83,183 | 41,380 |
Additional Information
How to Use The Datasets
Click to expand
Provide information on how researchers and developers can utilize and reference the datasets in their work.
Licensing and Citation
Click to expand
Include licensing details and citation instructions here.
Contribution and Feedback
Feel free to contribute or provide feedback to enhance the datasets. Guidelines on how to contribute and provide feedback can be detailed in this section.
References
[1] Alistair Plum, Tharindu Ranasinghe, Spencer Jones, Constantin Orasan, Ruslan Mitkov (2022). Biographical: A Semi-Supervised Relation Extraction Dataset. Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. [2] Alistair Plum, Tharindu Ranasinghe, Christoph Purschke (2024). Guided Distant Supervision for Multilingual Relation Extraction Data: Adapting to a New Language.