基于词汇范畴和语义相似的显性情感隐喻识别机制

隐喻是自然语言中比较常见的语言现象,在情感计算过程中有些句子的情感是由隐喻引起的,因此隐喻问题的解决将影响情感计算的结果.为此结合相关的隐喻理论,从机器学习的角度,对汉语文本中的显性隐喻的识别进行研究.以本体和喻体所属的范畴不同作为切入点,首先利用词性标注和依存句法分析提取句子中的本体词汇和喻体词汇,然后进行范畴划分和词汇语义相似度计算,最后使用支持向量机进行学习,从而对特定的隐喻句子进行识别.这一研究对后续的隐喻自动识别和隐喻理解起到了一定的作用.

Mechanism of dominant sentimental metaphor identification based on lexical domain and semantic similarity

Metaphor is a natural phenomenon of the universal language.In the process of affective computing,the feelings of some sentences are aroused by the metaphor.Therefore,sentimental metaphor will affect the outcome of affective computing.Based on the theory of metaphor,from the perspective of machine learning,dominant metaphor identification of Chinese text is studied.The difference of areas between target domain and source domain is regarded as breakthrough point.Firstly,target domain and source domain are got using part of speech tagging and the interdependence parsing analysis;then,areas classifying and word similarity calculation are given;lastly,support vector machine is used so as to identify metaphor of sentence.This study will play a certain role in the automatic identification of metaphor and metaphorical understanding.