摘要
本文以新浪微博用户(N=2143)为样本,采用BERT模型进行情感分析,考察用户随着信息茧房程度加深而呈现出怎样的共通情感偏向,及其基于情感偏向的情感趋同特征。研究发现:用户信息茧房程度越高,不同的信息茧房之间共通的情感偏向也越趋向增强;在32种情绪中,用惊喜、无奈、安心、想念、羡慕、愉悦、振奋、内疚、骄傲这9种情绪分别所占的比例即可预测用户的信息茧房程度,其调整后R^(2)达到0.489。在情感偏向基础上,随着用户信息茧房程度加深,用户情感与信息茧房“顶部”用户越来越相似。
Based on the“emotional bias”of microblogging users,we take the users on Sina Weibo(N=2143)as samples,use BERT model for emotion analysis,and investigate how common emotional bias and emotional convergence characteristics emerge as the degree of information cocooning deepens.It is found that the higher the degree of information cocoon,the greater the common emotional bias between different information cocoon rooms.Among 32 kinds of emotions,the proportion of surprise,helplessness,peace of mind,miss,envy,joy,excitement,guilt and pride can be used to predict the degree of information loss of users,and the adjusted R square reaches 0.489.On the basis of this emotional bias,as the degree of user information cocoon deepens,user emotion and the“top”user of information cocoon are more and more similar.
作者
徐翔
董洁芸
Xiang Xu;Jieyun Dong(College of Arts and Media,Tongji University)
出处
《全球传媒学刊》
CSSCI
2022年第4期78-99,共22页
Global Journal of Media Studies
基金
国家自然科学基金项目“社交网络互动中用户‘信息窄化’机理分析:基于微博的数据挖掘”(项目批准号:71804126)阶段性研究成果
关键词
社交网络用户
信息茧房
情感偏向
情感趋同
Social Network Users
Information Cocoon Room
Emotional Bias
Emotional Convergence
作者简介
徐翔,同济大学艺术与传媒学院教授;董洁芸,同济大学艺术与传媒学院本科生。
