基于文本挖掘及情感分析的社区负面舆论传播预测模型

Text Mining and Sentiment Analysis based Publicity Propagation Prediction Model

企业品牌舆论监控、网络敏感社区及重点社区识别是当前企业舆情监控的重点工作。作为网络社会的子集,不同的网络社区(社交媒体中联系密切的群体)由于社区网络结构的不同、社区成员情感倾向的不同,导致企业负面新闻在其中的传播会表现出来不同的特质。从网络社区的角度出发,研究不同社区情感倾向及社区网络结构下,企业负面新闻在其中产生的影响;进而提出了基于文本挖掘及情感分析的社区负面舆论传播预测模型。根据心理学测量视角Profile of Mood States(POMS)测度社区成员情感倾向(Tendency),以事件划分时间窗口;通过对连续六个月抓取的网络数据使用文本挖掘相关算法分析每个事件窗口内社区成员六种情感的分布(愤怒、紧张、失望等);在情感分布及网络结构上进行聚类,识别不同类别的情感倾向的网络社区;在些基础上建立社区情感倾向及舆论传播预测模型。测试结果表明:该模型在对网络社区情感倾向的识别及舆论传播倾向预测方面有较高的准确度,在舆论传播监控、敏感社区及重点社区识别等方面有一定的指导意义。

Abstract： Enterprise negative news which appears generally in the network society has caused a lot of negative reaction, but as a subset of the network society, negative news will show different characteristics within different online communities （members that are closely linked） because of the network structure of the different communities, different emotional tendencies of members. Trom the the view of network--communities this paper focus on emotional tendenc）es of different communities and community network structure, the enterprises in which the negative news impact, and proposed text mining--based sentiment analysis and community opinion propagation prediction model. Based on the measurement of psychological perspectives Profile of Mood States （POMS） , We measure emotional tendencies （Tendency） of community members, and we crawSed web comments and review for six consecutive months and use text mining aigrithms to analyze the six kinds of emotions within the window distribution （anger, stress, disappointment, etc.）. We use emotional distribution and network structure clustering to identify different types of emotional tendencies of online communities. Test results show that the model of the network community ident）fication and emotional tendencies prediction tend to have a high accuracy in the mass media monitoring, sensitive community and key community recognition, etc.