Multi-hop question answering (QA) requires an information retrieval (IR) system that can find \emph{multiple} supporting evidence needed to answer the question, making the retrieval process very challenging. This paper introduces an IR technique that uses information of entities present in the initially retrieved evidence to learn to `\emph{hop}' to other relevant evidence. In a setting, with more than \textbf{5 million} Wikipedia paragraphs, our approach leads to significant boost in retrieval performance. The retrieved evidence also increased the performance of an existing QA model (without any training) on the \hotpot benchmark by \textbf{10.59} F1.
@article{arxiv.1909.07598,
title = {Multi-step Entity-centric Information Retrieval for Multi-Hop Question Answering},
author = {Ameya Godbole and Dilip Kavarthapu and Rajarshi Das and Zhiyu Gong and Abhishek Singhal and Hamed Zamani and Mo Yu and Tian Gao and Xiaoxiao Guo and Manzil Zaheer and Andrew McCallum},
journal= {arXiv preprint arXiv:1909.07598},
year = {2019}
}