暴恐主题下微博用户特征及情感倾向性分析

摘要

The incidences of violence and terrorism have brought great challenge to national security and social administration, and in-depth research into network public opinion about violence and terrorism is one of the important academic fields. On the basis of sampling data of micro-blogs about violence and terrorism in Xinjiang, this paper summarizes the main characteristics of micro-bloggers statistically and depicts the dynamic path of micro-blog sentiment degrees. Furthermore, we construct the ordered multinomial logit model to explore the key factors influencing sentiment tendency. Empirical results show that gender, active degree, regional effect, fans affect emotional tendency significantly, while some interactions between the covariates offset the effect partly.%暴力恐怖事件给国家和社会安全管理带来很大的挑战，关于“暴恐”舆情的深入研究是网络舆情分析的重要内容之一。基于“新疆暴恐”主题下的微博抽样数据，对微博用户的基本特征进行了统计描述，刻画了舆情情感倾向强度的动态变化，并构建有序多分类Logistic回归模型，研究了影响网民情感倾向性的主要因素。实证结果表明网民的性别、活跃度、所在地区的环境效应对情感倾向有显著影响，但活跃度与其它因素的交互作用对影响效应有一定的抵消作用。