Chinese Question Classification from Approach and Semantic Views

This paper presents new Chinese question taxonomy respectively from approach and semantic viewpoints. According to the taxonomy, we propose a SVM classification algorithm which is based on multiple features. The experimental results show that: 1 Lexical semantic features and structural features are the guarantee of high performance of question classification; 2 The contribution of dependency relation extracted from our current parser is no better than that of Bi-gram, which is inconsistent with our anticipation. That is, if we cannot get high accuracy dependency parser, Bi-gram structural features can replace the dependency relation without performance declining. Finally, we examine the role of different features have on performance and analyze misclassifications via examples in detail.