摘要：This paper focuses on the indicator of professional authority for mining user proficiency within a specific topic domain, which contributes to Sina Weibo opinion leader identification. We construct an indicator system based on three dimensions comprising user attributes, interaction behaviors, and topic extraction. We quantify the professional authority indicator by applying the latent Dirichlet allocation topic model and information entropy. The grey relational analysis and the entropy-weight method are employed to calculate the opinion leader index and rank candidates accordingly, reflecting influence within a topic domain. And we validate the proposed method by analyzing a typical event of "Wei Ya's tax evasion" as an experimental case. Results demonstrate that the proposed model effectively identifies highly influential users with proficiency in this topic, these opinion leaders could play a guiding role in information diffusion. Moreover, it can serve as a reference for selecting opinion leaders in similar topic events as a prediction method. This paper supports regulatory agencies and Sina Weibo regarding online public opinion control, guidance, and response.
【本文第一作者为四川大学公共管理学院教授，博士生导师。主要研究方向为信息管理、信息系统、信息无障碍、智慧医养、信息用户行为研究等。】
