Multi-hop Question Generation with Graph Convolutional Network

TL;DR

This paper introduces MulQG, a novel multi-hop question generation model that uses graph convolutional networks to reason over multiple paragraphs, significantly improving question quality on the HotpotQA dataset.

Contribution

We propose the first multi-hop question generation model that performs reasoning over paragraphs without sentence-level info, using GCNs and encoding fusion.

Findings

Outperforms baselines on automatic metrics

Generates fluent and complete questions

Achieves 20.8% higher multi-hop evaluation score

Abstract

Multi-hop Question Generation (QG) aims to generate answer-related questions by aggregating and reasoning over multiple scattered evidence from different paragraphs. It is a more challenging yet under-explored task compared to conventional single-hop QG, where the questions are generated from the sentence containing the answer or nearby sentences in the same paragraph without complex reasoning. To address the additional challenges in multi-hop QG, we propose Multi-Hop Encoding Fusion Network for Question Generation (MulQG), which does context encoding in multiple hops with Graph Convolutional Network and encoding fusion via an Encoder Reasoning Gate. To the best of our knowledge, we are the first to tackle the challenge of multi-hop reasoning over paragraphs without any sentence-level information. Empirical results on HotpotQA dataset demonstrate the effectiveness of our method, in comparison with baselines on automatic evaluation metrics. Moreover, from the human evaluation, our proposed model is able to generate fluent questions with high completeness and outperforms the strongest baseline by 20.8% in the multi-hop evaluation. The code is publicly available at https://github.com/HLTCHKUST/MulQG}{https://github.com/HLTCHKUST/MulQG .