基于情感分析的社交网络意见领袖的识别——以汽车论坛为例

Identification of Opinion Leaders in Social Networks Based on Sentimental Analysis:Evidence from an Automotive Forum

社交网络的意见领袖在知识传播、舆情控制、商业营销等领域扮演重要角色。主题评论是用户意见的来源,情感分析有助于识别用户对意见领袖的情感倾向。为此,对用户言论主题分类,提取具有专业知识的主题帖。针对主题评论的情感极性不均衡分布情况,改进情感分类方法,最终构建关注、情感综合边权重矩阵。在此基础上,提出一种基于用户网络结构与情感倾向的Leader-PageRank算法。以汽车论坛为例,实验对比发现,Leader-PageRank能够结合社交网络的用户交互,有效识别在专业领域具有影响力的正面意见领袖。

The opinion leaders of social networks play an important role in fields of knowledge dissemination, public opinion control, marketing, etc. Topic review is an important source of user opinions and users＇ emotional tendency to opinion leaders can be identified through sentimental analysis. To begin with, this paper classifies the forum posts and extracts topic posts containing professional knowledge. Then the paper improves the sentimental analysis due to the imbalanced emotions reflected from topic review, and establishes a comprehensive attention and emotion weight matrix. Finally, the paper proposes a Leader-PageRank algorithm which is based on social network structure and emotional tendencies. Comparison experiments are performed on the automotive forum and results show that the Leader-PageRank method can combine user interactions in social networks and effectively recognize positive opinion leaders, who are influential in certain professional fields.