摘要
随着互联网的快速发展,网络中充斥着海量主观性文本,如何对这些主观性语句进行情感倾向性判断是文本情感分析的关键。本文提出一种基于词向量和句法树的中文句子情感分析方法。针对目前大量网络新词的使用所带来的问题,以已有标注的情感词典为基础,采用词向量的方法判断词语之间的语义相似度,从而得到未知词语的情感极性。针对情感极性转移现象,定义相应的情感判断规则。在此基础上,利用句子的句法树结构,对句子进行情感倾向性分析。实验证明,该方法在一定程度上解决了网络新词的问题,有效提高了句子情感分析的准确率和召回率,且具有领域适用性。
With the rapid development of Internet,the network is filled with a lot of subjective texts. How to judge the emotional polarity of these subjective statements is the key of the text sentiment analysis. In this paper,a method of sentiment analysis of Chinese sentences based on the word embedding and syntax tree structure is proposed. In view of the large number of network words,word embeddings are used to compute the semantic similarity between words,and the emotional polarity of the target word is gained. Some sentiment rules are defined for the phenomenon of emotional polarity transfer. Then,we judge the sentiment of sentences according to the syntax tree structure of the sentences. Experiments show that this method can solve the problem of the network words. Simultaneously,the precision and recall rate of the method are improved,and it also can be used widely in different domains.
出处
《计算机与现代化》
2016年第8期27-31,共5页
Computer and Modernization
关键词
情感词典
词向量
句法树
情感倾向性分析
sentimental lexicon
word embedding
syntax tree
sentiment analysis
