Event-Event Relation Extraction using Probabilistic Box Embedding
PROCEEDINGS OF THE 60TH ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL 2022): (SHORT PAPERS), VOL 2(2022)
摘要
To understand a story with multiple events, it is important to capture the proper relations across these events. However, existing event relation extraction (ERE) frameworks regard it as a multi-class classification task, and do not guarantee any coherence between different relation types. For instance, if a phone line died after storm, then it is evident that the storm happened before the died. Current frameworks of event relation extraction do not guarantee this anti-symmetry and thus enforce it via a constraint loss function (Wang et al., 2020). In this work, we propose to modify the underlying ERE model to guarantee coherence by representing each event as a box representation (BERE) without applying explicit constraints. Our experiments show that BERE has stronger conjunctive constraint satisfaction while performing on par or better in terms of F-1 compared to previous models with constraint injection.(1)
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络