HyperAIHyperAI
17 days ago

Asking Effective and Diverse Questions: A Machine Reading Comprehension based Framework for Joint Entity-Relation Extraction

{Zhao Yan, Zhoujun Li, Yunbo Cao, Tianyang Zhao}
Asking Effective and Diverse Questions: A Machine Reading Comprehension based Framework for Joint Entity-Relation Extraction
Abstract

Recent advances cast the entity-relation extractionto a multi-turn question answering (QA) task andprovide an effective solution based on the machinereading comprehension (MRC) models. However,they use a single question to characterize the meaning of entities and relations, which is intuitivelynot enough because of the variety of context semantics. Meanwhile, existing models enumerate all relation types to generate questions, whichis inefficient and easily leads to confusing questions. In this paper, we improve the existing MRCbased entity-relation extraction model through diverse question answering. First, a diversity question answering mechanism is introduced to detectentity spans and two answering selection strategiesare designed to integrate different answers. Then,we propose to predict a subset of potential relationsand filter out irrelevant ones to generate questionseffectively. Finally, entity and relation extractionsare integrated in an end-to-end way and optimizedthrough joint learning. Experiment results showthat the proposed method significantly outperformsbaseline models, which improves the relation F1to 62.1% (+1.9%) on ACE05 and 71.9% (+3.0%)on CoNLL04. Our implementation is available athttps://github.com/TanyaZhao/MRC4ERE.