Learning to Decompose Compound Questions with Reinforcement Learning

Haihong Yang, Han Wang, Shuang Guo, Wei Zhang, Huajun Chen

Abstract: As for knowledge-based question answering, a fundamental problem is to relax the assumption of answerable questions from simple questions to compound questions. Traditional approaches firstly detect topic entity mentioned in questions, then traverse the knowledge graph to find relations as a multi-hop path to answers, while we propose a novel approach to leverage simple-question answerers to answer compound questions. Our model consists of two parts: (i) a novel learning-to-decompose agent that learns a policy to decompose a compound question into simple questions and (ii) three independent simple-question answerers that classify the corresponding relations for each simple question. Experiments demonstrate that our model learns complex rules of compositionality as stochastic policy, which benefits simple neural networks to achieve state-of-the-art results on WebQuestions and MetaQA. We analyze the interpretable decomposition process as well as generated partitions.

Keywords: Compound Question Decomposition, Reinforcement Learning, Knowledge-Based Question Answering, Learning-to-decompose

TL;DR: We propose a learning-to-decompose agent that helps simple-question answerers to answer compound question over knowledge graph.

Data: [MetaQA](https://paperswithcode.com/dataset/metaqa), [SimpleQuestions](https://paperswithcode.com/dataset/simplequestions), [WebQuestions](https://paperswithcode.com/dataset/webquestions)