基于句子情感加权算法的篇章情感分析  

Document Sentiment Orientation Analysis Based on Sentence Weighted Algorithm

作　　者：李爱萍[1,2] 邸鹏[1] 段利国[1] 

机构地区：[1]太原理工大学计算机科学与技术学院,太原030024 [2]武汉大学软件工程国家重点实验室,武汉430072

出　　处：《小型微型计算机系统》2015年第10期2252-2256,共5页Journal of Chinese Computer Systems

基　　金：武汉大学软件工程国家重点实验室开放课题项目(SKLSE2012-09-30)资助;山西省自然科学基金项目(2013011015-2)资助

摘　　要：针对篇章级的文本情感倾向性分析准确率不高的问题,提出一种新的基于句子加权算法的篇章级文本情感倾向性分析方法.该方法将文本细化为句子的集合,在充分分析汉语中各种句子的句式结构特点的基础上,对复杂句采用新的情感判断规则判别情感,对简单句采用朴素贝叶斯算法判别情感,依据新提出的句子情感值的计算规则为每个句子赋予情感权值,最后依据此次研究提出的句子情感权值合成算法计算得出篇章级文本的情感倾向性.在实验中,使用本文所提出的方法对篇章级文本进行情感分析获得了较好的准确率、召回率和F值.实验证明,基于句子加权算法的情感分析方法要比传统的文本分类方法效率更高.Aiming at resolving the problem that the accuracy rate of document sentiment orientation analysis is not very high, this paper proposes a new method based on sentence weighted algorithm for document sentiment orientation analysis. In the new method, text is segmented into sentences. On the basis of analyzing structure features of different kinds of sentences sufficiently, the new method uses new sentiment judgment rules to determine the sentiment orientation of complex sentences and naive Bayesian algorithm to determine the sentiment orientation of simple sentences, assigns a sentiment weight to each sentence according to the calculation rule of sentence sentiment weight, and finally determines the sentiment orientation of document according to the composition algorithm of sentences sentiment weights. In experiment,using the new method this paper proposes to analyze the sentiment of documents has a better accuracy rate, recall rate and F-measure. Experiment proves that the new method for document sentiment orientation analysis based on sentence weighted algorithm has a higher efficiency than traditional text classification algorithm.

关 键 词：情感倾向性 复杂句 简单句 情感权值 句子加权算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]