基于语义角色标注的汉语句子相似度算法.pdf

1基于语义角色标注的汉语句子相似度算法**田堃,柯永红,穗志方(北京大学信息科学技术学院,北京市100871)摘要: 在语义角色标注过程中,经常需要检索相似的已标注语料,以便进行参考和分析。现有方法未能充分利用动词及其支配的成份信息,无法满足语义角色标注的相似句检索需求。基于此,本文提出一种新的汉语句子相似度计算方法。该方法基于已标注好语义角色的语料资源,以动词为分析核心,通过语义角色分析、标注句型的相似匹配、标注句型间相似度计算等步骤来实现句子语义的相似度量。为达到更好的实验效果,论文还综合比较了基于知网、词向量等多种计算词语相似度的算法,通过分析与实验对比,将实验效果最好的算法应用到句子相似度计算的研究中。实验结果显示,基于语义角色标注的句子相似度计算方法相对传统方法获得了更好的测试结果。关键词: 语义角色标注;词语相似度;知网;词向量;标注句型匹配中图分类号:TP391 文献标识码:AChinese Sentence Similarity ComputingBased on Semantic Roles AnnotationKun Tian,Yonghong Ke,Zhifang Sui(Peking University, Beijing, 100871, China)Abstract:In the process of semantic roles annotation, searching for similar annotated sentences is mon way to analyze such corpus. Existing methods cannot take full advantage of verbs and related elements, so they are unable to meet the demand of searching for similar annotated sentences. The article develops pletely newmethod to calculating Chinese sentence similarity focused on the verbs. Based on semantic roles annotation, the algorithm finds the similar sentences by analyzing the semantic roles, matching the annotated sentences, and calculating similarity between these matched sentences. To get a better result, the article pares several methods pute word similarity, including algorithms based on and Distributed Representation, and applies the algorithm that performs best to the algorithm through