Dataset and code for baselines for "DOCRED-FE: A DOCUMENT-LEVEL FINE-GRAINED ENTITY AND RELATION EXTRACTION DATASET"IEEE eXpress, arXiv
Joint entity and relation extraction (JERE) is one of the most important tasks in information extraction. However, most existing works focus on sentence-level coarse-grained JERE, which have limitations in real-world scenarios. In this pa- per, we construct a large-scale document-level fine-grained JERE dataset DocRED-FE, which improves DocRED with Fine-Grained Entity Type. Specifically, we redesign a hierar- chical entity type schema including 11 coarse-grained types and 119 fine-grained types, and then re-annotate DocRED manually according to this schema. Through comprehensive experiments we find that:
- DocRED-FE is challenging to existing JERE model.
- Our fine-grained entity types promote relation classification.
If you use the dataset or the code, please cite this paper:
@INPROCEEDINGS{10095786,
author={Wang, Hongbo and Xiong, Weimin and Song, Yifan and Zhu, Dawei and Xia, Yu and Li, Sujian},
booktitle={ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
title={DocRED-FE: A Document-Level Fine-Grained Entity and Relation Extraction Dataset},
year={2023},
volume={},
number={},
pages={1-5},
doi={10.1109/ICASSP49357.2023.10095786}}