基于社会网络分析的网络舆情潜在主题发现研究

发布时间：2018-07-16 21:50

【摘要】：互联网的迅速发展和广泛应用,使得网络舆情随之成为社会舆情最主要组成部分,微博所具备的社交网络特性和媒体传播特性使其成为了最具影响力的网络舆情衍生场所之一。现有研究多是根据网络舆情演化结果进行研究,一直处于被动的问题解决状态,无法充分满足对网络舆情的事前预警和事中实时发现需求。网络舆情潜在主题发现能够及时探测出网络舆情的核心内容,有助于正确把握网络舆情的产生规律和演化机制,对于构建良好的网络舆情环境具有重要意义。研究首先对已有网络舆情主题发现研究成果进行梳理分析,基于已有的少量潜在主题发现研究成果结合本研究目标对网络舆情潜在主题进行了定义;其次,对社会网络分析方法理论进行概述,着重对研究用到的社区发现方法和节点中心性方法原理进行说明分析;再次,对微博数据进行分析,包括微博内容,微博用户属性和行为数据,并梳理了与微博影响力相关研究理论方法,分析其与用户行为的关系。在此基础之上,研究构建网络舆情潜在主题发现模型,并对模型中用到的关键指标和方法进行说明:(1)基于微博用户行为构建用户行为关系网络,对不同用户行为和关注关系赋予特定权重;(2)利用社区发现方法对用户关系网络进行社区发现,并计算网络节点相关中心性指标;(3)计算重要社区中用户节点影响力并降序排列,筛选关键用户节点;(4)将社区关键用户节点映射到对应微博,获得关键微博节点;(5)通过TF-IDF方法对关键微博节点内容关键词排序,筛选出备选潜在主题词进行共词分析,获得潜在主题词集列表进行主题解读。最后,以"魏则西事件"作为研究案例,对模型效果进行实例验证,证实了本研究网络舆情潜在主题发现模型的有效性。
[Abstract]:With the rapid development and wide application of the Internet, network public opinion has become the most important part of social public opinion, and Weibo has become one of the most influential derivative places of network public opinion because of its social network characteristics and media dissemination characteristics. Most of the existing studies are based on the results of the evolution of network public opinion, which has been in a passive state of solving the problem, and can not fully meet the need of pre-warning and real-time discovery of network public opinion. The discovery of the potential topic of network public opinion can detect the core content of network public opinion in time, help to correctly grasp the law and evolution mechanism of network public opinion, and have important significance for the construction of good network public opinion environment. The research firstly combs and analyzes the existing research results on the topic discovery of network public opinion, and defines the potential topic of network public opinion based on a small number of existing research results combined with the objectives of this research; secondly, The social network analysis method theory is summarized, especially the community discovery method and node-centered method used in the research are explained and analyzed. Thirdly, the Weibo data, including Weibo content, are analyzed. Weibo user attributes and behavior data, and combing the theory and methods related to Weibo influence, and analyzing the relationship between Weibo and user behavior. On this basis, the potential topic discovery model of network public opinion is constructed, and the key indicators and methods used in the model are explained: (1) constructing user behavior relationship network based on Weibo user behavior; Give specific weight to different user behavior and relationship of concern; (2) use community discovery method to discover user relationship network and calculate relevant central index of network node; (3) calculate the influence and descending order of user node in important community. Screening key user nodes; (4) mapping community key user nodes to corresponding Weibo to obtain key Weibo nodes; (5) sorting key Weibo node content keywords by TF-IDF method, and selecting alternative potential theme words for coterm analysis. Get a list of potential theme words for topic interpretation. Finally, taking Wei Zexi incident as a case study, the effectiveness of the model is verified by an example, which verifies the validity of the model.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：C912.3;C913.4

【参考文献】