Learning to Order Sub-questions for Complex Question Answering

Yunan Zhang,Xiang Cheng,Yufeng Zhang,Zihan Wang,Zhengqi Fang,Xiaoyan Wang,Zhenya Huang,Chengxiang Zhai
DOI: https://doi.org/10.48550/arXiv.1911.04065
2019-11-12
Abstract:Answering complex questions involving multiple entities and relations is a challenging task. Logically, the answer to a complex question should be derived by decomposing the complex question into multiple simple sub-questions and then answering those sub-questions. Existing work has followed this strategy but has not attempted to optimize the order of how those sub-questions are answered. As a result, the sub-questions are answered in an arbitrary order, leading to larger search space and a higher risk of missing an answer. In this paper, we propose a novel reinforcement learning(RL) approach to answering complex questions that can learn a policy to dynamically decide which sub-question should be answered at each stage of reasoning. We lever-age the expected value-variance criterion to enable the learned policy to balance between the risk and utility of answering a sub-question. Experiment results show that the RL approach can substantially improve the optimality of ordering the sub-questions, leading to improved accuracy of question answering. The proposed method for learning to order sub-questions is general and can thus be potentially combined with many existing ideas for answering complex questions to enhance their performance.
Computation and Language,Artificial Intelligence
What problem does this paper attempt to address?