面向主题的微博意见领袖挖掘研究

Research on the Theme-oriented Mining of Microblog Opinion Leaders

【目的/意义】在微博中,意见领袖对于消息的传播以及舆情走向起着关键的作用。然而,现有的意见领袖研究大多忽略了意见领袖在特定话题下的意见代表性。【方法/过程】提出了面向主题的微博意见领袖研究的方法,该方法旨在利用LDA主题模型挖掘出特定话题中的各个主题,根据主题划分结果对参与话题讨论的用户进行分类,并结合用户自身属性,借鉴Page Rank算法思想挖掘出该主题演化中的意见领袖。【结果/结论】实验证明,本文方法更能体现特定话题的舆情走向,其针对主题挖掘的意见领袖也更具代表性。

【Purpose/significance】Microblog opinion leaders play a key role of the dissemination of news and public opinion trend. However, most of the opinion leader research ignored the representativeness of their opinion under specific topics.【Method/process】This paper presents a method of theme-oriented microblogging opinion leaders research. This method aims to use the LDA theme model to dig out the various topics in the specific topic, classify the users who participate in the topic discussion according to the theme division, and combine the user＇s own attributes to draw the opinion leader in the evolution of the topic.【Result/conclusions】The experiment proves that this method can reflect the trend of public opinion of specific topic, and the opinion leader of the topic mining is also more representative.