Complementary Evidence Identification in Open-Domain Question Answering
arXiv:2103.11643
Abstract
This paper proposes a new problem of complementary evidence identification for open-domain question answering (QA). The problem aims to efficiently find a small set of passages that covers full evidence from multiple aspects as to answer a complex question. To this end, we proposes a method that learns vector representations of passages and models the sufficiency and diversity within the selected set, in addition to the relevance between the question and passages. Our experiments demonstrate that our method considers the dependence within the supporting evidence and significantly improves the accuracy of complementary evidence selection in QA domain.
arXiv comment7 pages, EACL 2021
References in corpus (11)
- REALM: Retrieval-Augmented Language Model Pre-Training
- Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
- Evidence Aggregation for Answer Re-Ranking in Open-Domain Question Answering
- Learning to Retrieve Reasoning Paths over Wikipedia Graph for Question Answering
- HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering
- Reading Wikipedia to Answer Open-Domain Questions
- DocRED: A Large-Scale Document-Level Relation Extraction Dataset
- Efficient and Robust Question Answering from Minimal Context over Documents
- Multi-step Entity-centric Information Retrieval for Multi-Hop Question Answering
- Compositional Questions Do Not Necessitate Multi-hop Reasoning
- Do Multi-hop Readers Dream of Reasoning Chains?