基于评价系统的评论类文本情感倾向性分析  

A Text Sentiment Classification Method Based on Appraisal System

作　　者：张钫炜[1] ZHANG Fangwei

机构地区：[1]北京邮电大学,北京100876

出　　处：《语言文字应用》2018年第2期138-144,共7页Applied Linguistics

基　　金：教育部人文社科基金项目"基于评价理论的多维情感语料库建设及其细粒度情感语义研究"(14YJA74006)的资助

摘　　要：本文引入功能语言学的评价系统理论作为文本情感倾向分类的框架。这一框架将人类情感系统化为不同的范畴,并涉评价者、评价对象、评价范畴、语境等不同要素,提高了情感分类的精密度和复杂性。它提供了语言学与自然语言处理相结合的一个新的研究方向。本研究首先运用评价系统对训练文本进行分析和标注,进而构建情感词库,然后采用分类算法中的支持向量机(SVM)来实现对测试文本情感倾向的分析。The Appraisal System is adopted in the text sentiment classification. The system categorizes human feelings in affect, judgement, and appreciation and also considers the appraiser, the appraised, and context, which can improve the complexity and accuracy of sentiment analysis. The paper introduces a new approach for the cooperation of linguistics and natural language processing. In this experiment, training texts are analyzed and annotated based on Appraisal System, based on which emotional lexicon is established. And SVM （support vector machine) is used to achieve the sentiment analysis of training texts.

关 键 词：文本情感分析 评价系统 支持向量机 

分 类 号：H08[语言文字—语言学]