基于网络语料的特定领域情感词典生成方法研究  

Research on Domain-specific Sentiment Lexicon Generation Method Based on Web Corpus

作　　者：张志武[1,2] 陈国兰[1,2] Zhang Zhiwu;Chen Guolan(Library of Nanjing University of Posts and Telecommunications,Nanjing Jiangsu 210023;Institute of Scientific and Technological Information of Nanjing University of Posts and Telecommunications,Nanjing Jiangsu 210023)

机构地区：[1]南京邮电大学图书馆,江苏南京210023 [2]南京邮电大学科技情报研究所,江苏南京210023

出　　处：《情报探索》2019年第4期1-6,共6页Information Research

基　　金：江苏省高校哲学社会科学研究项目"基于商品评论情感挖掘的网络消费舆情倾向性分析研究"(项目编号:2015SJD129)成果

摘　　要：[目的/意义]旨在为构建特定领域情感词典提供参考。[方法/过程]基于同义词词典和基于语料库的特点,提出多元融合的词语关系图构建方法,利用半监督标签传播算法为未标记的词语分配情感极性,并进行实例分析。[结果/结论]该方法在自动生成特定领域词汇上具有优越性。[Purpose/significance] The paper is to provide references for construction of domain-specific sentiment lexicon.[Method/process] Based on the characteristics of synonym dictionary and corpus,this paper proposes a multi-fusion method for constructing the relationship graph of words,uses the semi-supervised label propagation algorithm to assign the sentiment polarity to unlabeled words,and conducts an instance analysis.[Result/conclusion] The method has advantages in automatic generation of domain-specific vocabulary.

关 键 词：情感分析 特定领域词典 多元融合 网络语料 

分 类 号：G350[文化科学—情报学]