Title2Event is a open event extraction dataset with large-scale human annotated Chinese title. Title2Event contains more than 42,000 news titles in 34 topics collected from Chinese web pages. It is collected by researcher at Harbin Institute of Technology and QQ Browser Search.
For more details, please refer to our EMNLP 2022 paper:
(deng-etal-2022-title2event)Title2Event is distributed under a CC BY-SA 4.0 License. The dataset can be obtained below:
Baidu NetdiskFor the baseline codes, please refer to our github repository.
baseline repoIf you want your results to be appeared on the official leaderboard here, please read the guideline following.
Leaderboard GuidelineIf you use Title2Event in your research, please cite our paper.
@inproceedings{deng-etal-2022-title2event, title = "{T}itle2{E}vent: Benchmarking Open Event Extraction with a Large-scale {C}hinese Title Dataset", author = "Deng, Haolin and Zhang, Yanan and Zhang, Yangfan and Ying, Wangyang and Yu, Changlong and Gao, Jun and Wang, Wei and Bai, Xiaoling and Yang, Nan and Ma, Jin and Chen, Xiang and Zhou, Tianhua", booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing", month = dec, year = "2022", address = "Abu Dhabi, United Arab Emirates", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.emnlp-main.437", pages = "6511--6524", abstract = "Event extraction (EE) is crucial to downstream tasks such as new aggregation and event knowledge graph construction. Most existing EE datasets manually define fixed event types and design specific schema for each of them, failing to cover diverse events emerging from the online text. Moreover, news titles, an important source of event mentions, have not gained enough attention in current EE research. In this paper, we present Title2Event, a large-scale sentence-level dataset benchmarking Open Event Extraction without restricting event types. Title2Event contains more than 42,000 news titles in 34 topics collected from Chinese web pages. To the best of our knowledge, it is currently the largest manually annotated Chinese dataset for open event extraction. We further conduct experiments on Title2Event with different models and show that the characteristics of titles make it challenging for event extraction, addressing the significance of advanced study on this problem. The dataset and baseline codes are available at https://open-event-hub.github.io/title2event.", }
Methods | Trigger Ex. | Argument Ex. | Triplet Ex. | ||||||
---|---|---|---|---|---|---|---|---|---|
Precission | Recall | F1 | Precission | Recall | F1 | Precission | Recall | F1 | |
EventGLM_gwn |
70.4 | 70.7 | 70.5 | 58.5 | 58.3 | 58.4 | 50 | 50.2 | 50.2 |
ST-Seq2SeqMRC |
- | - | - | 57.9 | 58.6 | 58.2 | 49.8 | 50.1 | 49.9 |
ST-SpanMRC |
- | - | - | 60.1 | 54.9 | 57.4 | 44.5 | 44.8 | 44.7 |
SeqTag |
69.5 | 69.9 | 69.7 | 50.8 | 51.2 | 51 | 41.1 | 41.3 | 41.2 |
Unsuper |
21 | 32 | 25.4 | 12 | 15.5 | 13.5 | 4.5 | 6.8 | 5.4 |