Sen Hu's paper on Knowledge Base Question Answering (A State-transition Framework to Answer Complex Questions over Knowledge Base) is accepted by EMNLP 2018.

This paper analyze the properties and challenges of complex questions and study how to sovle complex question answering. It first translate the complex question to semantic query graph and then utilize the existing subgraph matching algorithm to find the answers. This paper introduce a state-transition based framework to answer complex questions by utilizng four primitive operations to build query graph leding by a greedy search algorithm, which relies on a pre trained liner ranking model. The experiment results show that our solution performs better than the state-of-art systems.