Single Sequence Prediction over Reasoning Graphs for Multi-hop QA

TL;DR

This paper introduces a novel single-sequence prediction method over a local reasoning graph for multi-hop QA, improving answer accuracy and reasoning faithfulness by integrating entity-based graph structures into generative models.

Contribution

It proposes a graph neural network-based approach that encodes entity relationships to enhance multi-hop question answering models' interpretability and accuracy.

Findings

Significant improvements in answer exact-match and F1 scores.

Enhanced faithfulness of reasoning paths.

State-of-the-art results on the Musique dataset.

Abstract

Recent generative approaches for multi-hop question answering (QA) utilize the fusion-in-decoder method~\cite{izacard-grave-2021-leveraging} to generate a single sequence output which includes both a final answer and a reasoning path taken to arrive at that answer, such as passage titles and key facts from those passages. While such models can lead to better interpretability and high quantitative scores, they often have difficulty accurately identifying the passages corresponding to key entities in the context, resulting in incorrect passage hops and a lack of faithfulness in the reasoning path. To address this, we propose a single-sequence prediction method over a local reasoning graph (\model)\footnote{Code/Models will be released at \url{https://github.com/gowtham1997/SeqGraph}} that integrates a graph structure connecting key entities in each context passage to relevant subsequent passages for each question. We use a graph neural network to encode this graph structure and fuse the resulting representations into the entity representations of the model. Our experiments show significant improvements in answer exact-match/F1 scores and faithfulness of grounding in the reasoning path on the HotpotQA dataset and achieve state-of-the-art numbers on the Musique dataset with only up to a 4\% increase in model parameters.