基于多种情感特征的网络文本倾向性判别方法研究  

Research on the Method of Network Text Orientation Discrimination Based on Multiple Sentiment Features

作　　者：樊康新 FAN Kang-xin （School of Computer Science and Technology, Nantong University, Nantong 226019, China）

机构地区：南通大学计算机科学与技术学院,江苏南通226019

出　　处：《电脑知识与技术》2015年第8期18-21,共4页Computer Knowledge and Technology

基　　金：南通大学自然科学基金资助项目（11Z075）

摘　　要：网络文本的情感倾向性分析对于电子商务、网络信息安全、网络舆情等方面具有非常重要的意义。本文在对文本倾向性分析的常用方法作了分析和研究的基础上,提出了一种综合情感词语、否定词、程度副词、关联词和词句类型等多种特征计算词句的极性值,进而判别文本情感倾向性的方法。实验结果表明,与ku提出的算法相比,该方法更能有效地判定文本情感倾向性。The analysis of the sentiment orientation of the network text is very important for the electronic commerce, the network information security, the network public opinion and so on. In this paper, the general method of text orientation analysis is studied. We propose a new method to calculate the polarity value of the words and to judge the sentiment orientation of text, which comprehensives a variety of characteristics, such as the emotional words, negative words, adverbs of degree, related words, and so on. The experimental results show that the proposed method is more effective than the Ku algorithm for judging sentiment orientation.

关 键 词：情感词典 情感特征 网络文本 文本倾向性 倾向性分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]