Journal of Frontiers of Computer Science and Technology ›› 2021, Vol. 15 ›› Issue (11): 2077-2092.DOI: 10.3778/j.issn.1673-9418.2104013

• Surveys and Frontiers • Previous Articles     Next Articles

Survey on Methods of Opinion Leader Mining in Social Networks

GUO Yi, XU Liang, XIONG Xuejun   

  1. School of Electrical Engineering and Electronic Information, Xihua University, Chengdu 610039, China
  • Online:2021-11-01 Published:2021-11-09

社交网络中意见领袖挖掘方法综述

郭奕徐亮熊雪军   

  1. 西华大学 电气与电子信息学院,成都 610039

Abstract:

Opinion leaders are those who have strong influence on the public in the process of communication, directly or indirectly influence public opinion tendency and formation. Opinion leader mining in social networks is a significant task, which has been widely used in the fields of commercial marketing, policy propaganda, public opinion monitoring, and social public issues. Firstly, the origin, definition and classification of opinion leaders are described. Then a more comprehensive summary of current opinion leader mining methods is presented, which is grouped into four categories: user rating rule-based methods, social network graph-based methods, influence propagation model-based methods, and multidimensional fusion methods. The basic ideas and key points of the above methods are described respectively, and the advantages and disadvantages of each method are analyzed. In addition, the recommended evaluation metrics are given by analyzing the existing evaluation metrics. Finally, three future research directions are discussed: improving mining efficiency and effectiveness through clustering mining methods using graph neural network, designing dynamic models to meet more time-sensitive application scenarios, and dividing opinion leader hierarchies to meet different levels of demand.

Key words: social network, opinion leader, rating rules, influence communication model, graph neural network

摘要:

意见领袖是指在消息传播过程中,对大众具有较大的引导力和影响力,直接或间接地影响大众观点的倾向与形成的那些人。社交网络中的意见领袖挖掘在商业营销、政策宣传、舆情监控、社会公共问题等领域有着非常重要的应用价值。首先阐述了意见领袖的起源、定义和分类。随后较为全面地总结了当前的意见领袖挖掘方法,将其归结为四类:基于用户评分规则的方法、基于社交网络图的方法、基于影响传播模型的方法、多维度融合的方法。分别阐述了上述方法的基本思想及其关键技术,分析了各种方法的优缺点。此外,对已有的评价指标进行分析总结,给出了推荐使用的评价指标。最后,探讨了三个未来的研究方向:使用图神经网络的聚类挖掘方法以提高挖掘效率和效果,设计动态模型以满足对时间更加敏感的应用场景,划分意见领袖等级以满足不同层次需求。

关键词: 社交网络, 意见领袖, 评分规则, 影响传播模型, 图神经网络