摘要
通过挖掘网络舆情事件话题演化,有助于在事件全面爆发之前的更早阶段釆取应急措施。本文基于网络舆情事件时间性强、话题与时间关联度单一的特征,选择网民关注度高的食品安全网络舆情事件新闻报道建立文本集,基于LDA模型抽取话题,使用后离散时间型话题模型思路分析话题热度变化,用先离散时间型话题模型思路分析话题内容迁移。实验表明,此思路能够较全面体现话题演化路径,为网络舆情事件分析提供有效途径。
By exploring the event topic evolution of network public opinions, it may help us to take effective measures timely in dealing with emergency cases. Since network public opinion of event is time-sensitive and the topic is closely associated with time, we selected network public opinion of food safety incidents with high concern among netusers and extracted topics based on LDA model. The time post -discretized method was used to analyze the change of topic heat level and the time pre-discretized method was applied to analyze the change of topic content. The proposed method was experimentally verified to be efficient for detecting the event topic evolution of network public opinions.
出处
《情报杂志》
CSSCI
北大核心
2013年第12期26-30,共5页
Journal of Intelligence
基金
国家自然科学基金"基于情景演化的数字化应急预案动态生成机制研究"(编号:71171117)
江苏省教育厅哲学社会科学项目"基于环境演化的食品供应链安全风险转移机制研究"(编号:2012SJB630051)
关键词
网络舆情
LDA模型
后离散时间型
先离散时间型
特征词
话题演化
network public opinion LDA model post-discretized method pre-discretized method feature word topic evolution
