基于SVM的疑问句问点语义角色标注

SVM(SupportVectorM ach ine)是一种基于结构风险最小化原则的分类算法,由于其完善的理论基础使其在小样本模式识别中表现出比其他算法更好的泛化能力。语义角色标注是一种浅层语义分析的方法。为了加深对问句的理解,对问句的问点进行语义角色标注是必要的。故将问点的语义角色标注视为分类问题,并提出了一种基于SVM的问点语义角色标注方法。在哈工大标准问句集上进行实验,取得了语义角色标注封闭测试91.4%,开放测试71.6%的正确率。实验结果表明本文所采用的方法是有效的。

Question query focus based on SVM semantic role labeling

SVM(Support Vector Machine) is a structural risk minimization principle based on classification algorithm,and it has achieved higher generalization performance with small number of samples than other classification algorithms due to its perfect theoretical properties.Semantic Role Labeling is a shallow semantic parsing method.In order to deepen the understanding of a question sentence,it is necessary to label the role of the query focus of the question.We regard the focus labeling as a classification problem,and propose a SVM based on labeling method for it.Experiment on HIT's(Harbin Institute of technology) question collection,we achieve the 91.4% precision in close test and 71.6% in open test respectively.The result shows that the method is effective.