Propagate-Selector: Detecting Supporting Sentences for Question Answering via Graph Neural Networks

TL;DR

This paper introduces propagate-selector, a graph neural network that propagates information across sentences to improve question answering by capturing intersentential relationships, outperforming existing models on HotpotQA.

Contribution

The paper presents a novel graph neural network with a unique propagation and aggregation method for better understanding of sentence relationships in QA tasks.

Findings

Achieved state-of-the-art performance on HotpotQA dataset.

Outperformed traditional answer-selection models without intersentential reasoning.

Demonstrated the effectiveness of sentence-level graph propagation.

Abstract

In this study, we propose a novel graph neural network called propagate-selector (PS), which propagates information over sentences to understand information that cannot be inferred when considering sentences in isolation. First, we design a graph structure in which each node represents an individual sentence, and some pairs of nodes are selectively connected based on the text structure. Then, we develop an iterative attentive aggregation and a skip-combine method in which a node interacts with its neighborhood nodes to accumulate the necessary information. To evaluate the performance of the proposed approaches, we conduct experiments with the standard HotpotQA dataset. The empirical results demonstrate the superiority of our proposed approach, which obtains the best performances, compared to the widely used answer-selection models that do not consider the intersentential relationship.