基于文本挖掘的网络舆情情感倾向及演化分析

发布时间：2018-06-07 06:23

本文选题：网络舆情 + 情感倾向　；参考：《湘潭大学》2017年硕士论文

【摘要】：随着移动互联网的快速发展,社交网络已经成为用户获取信息、表达意见、交流看法的重要平台。热点事件一旦发生后,网络用户可以通过文本、图片、小视频等方式表达自己对某个社会事件的态度、认知、意见和情感等主观性信息。信息经过转发、评论和点赞等方式进行传播,同时若用户在转发与评论信息时加入个人主观性情感,从而促进了事件的演化。近年来,网络群体性事件数量急剧上升,在网民中引起了巨大的舆论反响,当突发事件爆发时若不对不良情感进行控制和引导,舆论则很容易极端化,甚至危及社会安全与稳定。因此,有必要面向网络舆情进行用户情感倾向性分析研究,为政府有效掌握和监控网络舆情突发事件提供相应的理论支持和对策建议。本文以“罗一笑”网络热门话题事件为例,对舆情信息进行情感分析和舆情追踪。主要的研究工作包括:第一,利用网络爬虫工具采集事件相关微博数据,并进行整理分析。第二,以知网HowNet等词典为基础对情感词进行扩展,构建一个比较全面的情感分类词典,同时对各情感词所表达的情感极性和强度进行识别和标记。第三,构建情感倾向分析模型,判断网络舆情的情感类型和统计情感词频,并对该事件中的用户情感进行挖掘与可视化分析。第四,运用实证分析研究,对该事件的舆情演化阶段进行划分,分别对各阶段用户情感演化特征及规律进行分析。为后续网络舆情情感引导对策的提出提供参考依据。实验表明,网络舆情从生成到最终消亡是一个完整的生命周期,通过对网络舆情演化进行科学的阶段划分,可以发现各阶段特征:(1)开始期微博发布数量少,网民对网络舆情事件的态度纷繁复杂,但是通过对文本中用户情感的挖掘、观点的提取有利于进一步跟踪事件的后续发展趋势;(2)爆发期微博发布数最多,用户参与度最高,影响范围和影响效果极大,网民对事件的态度、观点、情感等信息能够为网络舆情分析和监控提供足量的数据基础,同时,爆发期的情感倾向很大程度上定义了网络舆情事件的总体情感演化趋势,相关部门应对爆发期的网络舆情情感演化多加关注,并引导舆情朝着正确的方向发展;(3)发酵期网民对网络舆情事件的新资讯、新动态较为敏感,正面信息公开与舆情披露在此阶段能够起到良好的效果;(4)消解期和反思期用户参与程度较低,但仍需要对网络舆情事件进行跟踪报道,规避谣言,肃清网络环境,避免网络舆情事件的二次发酵。
[Abstract]:With the rapid development of mobile Internet, social network has become an important platform for users to obtain information, express opinions and exchange views. Once a hot event occurs, Internet users can express their attitude, cognition, opinion and emotion on a social event by means of text, picture, small video and so on. The information is transmitted by way of forwarding, commenting and liking, and if the user adds personal subjective emotion to transmit and comment on the information, it promotes the evolution of the event. In recent years, the number of network mass incidents has risen sharply, causing a huge public opinion response among Internet users. When emergencies break out, if they do not control and guide bad emotions, public opinion is easy to become extreme. Even endanger social security and stability. Therefore, it is necessary to analyze and study the emotional tendency of users in order to provide corresponding theoretical support and countermeasures for the government to effectively grasp and monitor the sudden events of network public opinion. Taking Luo Yixiao as an example, this paper analyzes and tracks public opinion information. The main research work is as follows: firstly, the Weibo data are collected and analyzed by using web crawler tools. Secondly, based on the HowNet dictionary, we construct a comprehensive emotion classification dictionary, and identify and mark the emotion polarity and intensity expressed by each emotion word. Thirdly, we construct an emotional tendency analysis model to judge the emotional types and statistical affective word frequency of network public opinion, and mine and visualize the user emotion in this event. Fourthly, using the empirical analysis, the public opinion evolution stage of the event is divided, and the characteristics and rules of user emotion evolution in each stage are analyzed respectively. It provides the reference for the following network public opinion emotion guidance countermeasure. The experiment shows that the network public opinion is a complete life cycle from the generation to the final extinction. By dividing the evolution of the network public opinion into scientific stages, we can find that the number of Weibo releases at the beginning of each stage is small. Internet users' attitude to network public opinion events is complicated, but through the mining of users' feelings in the text, the point of view extraction is conducive to further tracking the future development trend of events.) in the outbreak period, the number of Weibo releases is the most, and the participation of users is the highest. The influence range and the influence effect are great, the netizens' attitude, viewpoint, emotion and so on information can provide the sufficient data foundation for the network public opinion analysis and the monitoring, at the same time, The emotional tendency of the outbreak period has largely defined the overall emotional evolution trend of the network public opinion event, and the relevant departments should pay more attention to the evolution of the network public opinion emotion in the outbreak period. And guide the public opinion to develop in the right direction.) during the fermentation period, the netizens are more sensitive to the new information about the network public opinion events. Positive information disclosure and public opinion disclosure can play a good effect in this stage. (4) the level of user participation in the period of resolution and reflection is relatively low, but it is still necessary to track and report online public opinion events, to avoid rumors, and to eliminate the network environment. Avoid the secondary fermentation of network public opinion events.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：C912.63

【参考文献】