中文微博的情感分析和应用

发布时间：2018-03-13 05:36

本文选题：微博　切入点：情感词典　出处：《南京邮电大学》2014年硕士论文　论文类型：学位论文

【摘要】：近年来微博已逐渐成为了一种主流的消息发布渠道。大量网络用户通过微博快速及时的传递信息、发表意见、表达情感。网络热点事件、热点话题等都可以通过微博快速传播并可以通过微博迅速收集网络民众对此的看法。正因如此，近年来，针对用户在微博中所表露的情感进行分析，已成为一个新的研究热点。情感分析主要包含三个方向的情感倾向，即积极倾向、中立倾向和消极倾向。常用的情感分析方法包括基于机器学习和基于情感词典的方法，本文采取基于情感词典的中文微博情感分析方法。主要工作包括：利用中文微博的自身特点对微博进行预处理，采用基于多种传统方法的组合型分词方法，针对分词过程中经常出现的交集型歧义提出解决方法，提取出微博中的有效情感元素；对微博中有效情感元素中的情感词、程度副词、否定词、表情符号等进行相应的情感分析处理，进而得到整个微博的情感极性和情感强度。考虑到中文微博中存在网络反语的现象，本文基于已得到的微博情感极性和情感强度作进一步的情感分析，提出了基于MMTD（Measure of MediumTruth Degree）的微博情感分析方法，对每条微博作出最终的情感倾向判断。本文从新浪微博热点问题中选取了8个微博热门话题，分别涉及到不同的社会热点问题，采用本文提出的基于MMTD的分析方法对微博情感倾向性进行分析，，实验结果表明本文方法能够发现微博中存在的反语，有助于提高微博情感倾向性判断的准确性。
[Abstract]:In recent years, Weibo has gradually become a mainstream channel for news release. A large number of Internet users pass information, express their opinions, express their feelings, and express their feelings quickly and in a timely manner through Weibo. Hot topics and other topics can be quickly spread through Weibo and online public views on this can be quickly collected through Weibo. Therefore, in recent years, the analysis of the emotions revealed by users in Weibo has become a new research hotspot. Affective analysis mainly includes three kinds of affective tendencies, that is, positive tendency, neutral tendency and negative tendency. The commonly used affective analysis methods include machine learning and affective dictionary based methods. This paper adopts the affective dictionary based Chinese Weibo emotion analysis method. The main work includes:. Taking advantage of Weibo's own characteristics, we pretreat Weibo, adopt combinatorial word segmentation method based on many traditional methods, and propose solutions to the overlapping ambiguity that often occurs in the process of word segmentation. The effective emotional elements in Weibo and the affective words, degree adverbs, negative words, emoji and emoticons in the effective emotional elements of Weibo were analyzed and processed accordingly, and then the emotional polarity and intensity of the whole Weibo were obtained. Considering the phenomenon of network irony in Chinese Weibo, this paper gives a further emotional analysis based on the obtained affective polarity and intensity of Weibo, and puts forward an affective analysis method based on MMTD(Measure of MediumTruth grave. Make the final judgment on each Weibo's emotional tendency. This paper selects 8 hot topics from the hot topics of Sina Weibo, which involve different social hot issues. The analysis method based on MMTD proposed in this paper is used to analyze Weibo's emotional tendency. The experimental results show that the method in this paper can find the irony existing in Weibo and help to improve the accuracy of Weibo's emotional tendentiousness judgment.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.1;TP393.092

【参考文献】