摘要
互联网引发的舆情问题愈发突出,网络舆情研究已被深度关注.话题演化是网络舆情分析的重要内容之一,本文尝试从话题热度变化和内容变化两方面研究舆情动态.本文选取天涯论坛民生讨论的主要版块——天涯杂谈的首发帖为舆情来源,分析比较一系列主题模型后,建立动态主题模型(DTM).通过挖掘随时间变化的动态话题链,从词语变化的微观角度分析热门事件下公众意见的变迁过程,还原事件的整个发展过程.本文提出话题热度计算方法,通过计算2012全年天涯杂谈版块下所有新发帖的话题热度值变化及可视化分析,总结了BBS话题的三个规律.
Along with the increasingly prominent issues aroused via Internet, online opinions are more and more concerned by researchers. This paper attempts to study the evolution of online opinions from two aspects, hot degree and BBS content, respectively. The original posts in Tiaya Zatan board, which is a primary board on social topics in Tianya Club, are collected as a source of online public opinion. By comparisons of a series of topic models, the dynamic topic model （DTM） is chosen. After mining the dynamic topics, the evolutions of public opinions toward hot events are analyzed under a micro perspective of changing words, so as to outline the process of those events. A method is proposed to compute the hot degree of the topic, and then the hot de- gree of all topics on the Tianya Zatan board in 2012 are computed and visualized, from which three rules of the topics are summarized under a macro perspective.
出处
《管理科学学报》
CSSCI
北大核心
2014年第11期109-121,共13页
Journal of Management Sciences in China
基金
国家重点基础研究发展计划资助项目(2010CB731405)
国家自然科学基金资助项目(71171187
71371107)
关键词
主题模型
DTM
话题演化
天涯论坛
topic models
dynamic topic model
topics evolution
Tianya club
