Chinese question classification combining syntactic and semantic feature

Recognition of question type is the keys to a question answering system. At present, Chinese question classification is based on the combinatorial rules between the interrogative and its interrelated words. However, because the extraction of rules greatly dependeds on expert's experience knowledge, the classification effect will be impacted. In this paper, the Chinese question types are defined; a question classification machine learning model based on SVM is established; the course of selecting feature and extracting feature using SVM for question classification is described in detail; question classification experiment is done by augmenting respectively semantic feature on the baseline of syntactic feature; experiment have achieved accuracy of 88.6%; This result demonstrate that combining syntactic feature and semantic feature has a very good effect on Chinese question classification.