Multi-step Entity-centric Information Retrieval for Multi-Hop Question\n Answering

2019/09/17 by Ameya Godbole, Godbole, Ameya, Dilip Kavarthapu +20 · 1 citation
Computer Science · #Computation and Language (cs.CL) #Expert finding and Q&A systems #FOS: Computer and information sciences #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1909.07598

openalex publication_date 2019/09/17 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28

Abstract

Multi-hop question answering (QA) requires an information retrieval (IR)\nsystem that can find \multiple supporting evidence needed to answer the\nquestion, making the retrieval process very challenging. This paper introduces\nan IR technique that uses information of entities present in the initially\nretrieved evidence to learn to `\hop' to other relevant evidence. In a\nsetting, with more than \5 million Wikipedia paragraphs, our approach\nleads to significant boost in retrieval performance. The retrieved evidence\nalso increased the performance of an existing QA model (without any training)\non the hotpot benchmark by \10.59 F1.\n

Cited by

• Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning

Related

• PRISM: Agentic Retrieval with LLMs for Multi-Hop Question Answering
• HopRetriever: Retrieve Hops over Wikipedia to Answer Complex Questions
• Answering Complex Open-domain Questions Through Iterative Query Generation
• Memory-Aware and Uncertainty-Guided Retrieval for Multi-Hop Question Answering
• A Comprehensive Evaluation of Transformer-Based Question Answering Models and RAG-Enhanced Design