文本情感倾向性分析方法:bfsmPMI-SVM

发布时间：2018-05-07 17:33

  本文选题：文本情感倾向性分类 + 互信息　； 参考：《武汉大学学报(理学版)》2017年03期

【摘要】：为了提高文本情感倾向性分类的精度,提出了一种文本情感倾向性分析方法 bfsmPMI-SVM.该方法在文本预处理阶段,滤除了对表述主题情感倾向性不强烈的语句以及无关停用词等;用改进的PMI-IR算法对情感倾向性词语抽取,并自动扩充了正负基准词集;改进了互信息(MI)算法,在MI的计算中增加了词频因子(f)、类别差异因子(b)和符号因子(s).利用改进的MI算法选择文本特征,融合其他一些文本特征,用SVM实现文本情感倾向性分类.实验以食品安全领域爬取文本为例,与PMI-IR-SVM和MI-SVM算法的倾向分析相比,本文方法的正向文本准确率、负向文本准确率、召回率和F1值等都有提高.
[Abstract]:In order to improve the accuracy of text affective orientation classification, a text affective orientation analysis method bfsmPMI-SVM is proposed. In the stage of text preprocessing, the method filters out sentences which are not strong in the tendency to express the subject emotion, and uses the improved PMI-IR algorithm to extract the affective preference words, and automatically expands the positive and negative reference words set. The mutual information Mi) algorithm is improved, and the word frequency factor, category difference factor and symbol factor are added in the calculation of MI. The improved MI algorithm is used to select text features and some other text features are fused, and SVM is used to realize text affective orientation classification. The experiment takes crawling text in the field of food safety as an example. Compared with the tendency analysis of PMI-IR-SVM and MI-SVM algorithms, the forward text accuracy, negative text accuracy, recall rate and F1 value of this method are improved.
【作者单位】： 武汉大学计算机学院;武汉大学国际软件学院;
【基金】：国家自然科学基金资助项目(61303214,61672393,U1536204)
【分类号】：TP391.1
，

本文编号：1857823