中文摘要：

在Twitter中是否存在情绪传染现象是社会科学中一个待解决的问题。首先通过LIWC2007获取了包含106 641个用户的Tw itter社交网络中所有用户的情绪时间序列,然后采用一系列的单位根检验证明了相关时间序列的平稳性,通过格兰杰因果检验,在预测用户情绪值的回归式中加入了用户关注好友过去时间的情绪值作为自变量,并采用统计假设检验的方法证明了该自变量的系数不为0,从而说明了用户的情绪会显著地被其关注好友过去的情绪所影响,即用户关注好友的情绪是用户情绪的格兰杰原因。用同样的方法证明了用户情绪并不是用户关注好友情绪的格兰杰原因,由于社交选择现象是一种双向关系,所以该结果是由情绪传染现象造成的。此外,通过统计分析发现Twitter中绝大部分单向关注好友都是非熟人,而绝大部分双向关注好友都是熟人。格兰杰因果检验的结果说明了人们的情绪既会被熟人的情绪所传染,也会被非熟人的情绪所传染。

英文摘要：

It is a problem to be solved in social science whether there exists the phenomenon of emotional contagion in Twitter. The emotion time series of 106 641 users in the Twitter social network were got by the LIWC2007. Then a serious of unit root tests were used to validate that the time series are stable. Through the Granger causality test,the emotion variable of the users＇ followees in the past time was added to the regression equation to predict the users＇ emotion and then the statistical hypothesis tests was used to prove the regression coefficient of the variable was significantly not equal to 0,which indicated that the users＇ emotion could be influenced by their followees＇ emotional expression in the past time,which meaned that the emotion expressed by users＇ followees was the Granger cause of the emotion expressed by users. At the same time,the same method proved that the emotion expressed by users was not the Granger cause of the emotion expressed by users followees. Since the social selection was a type of bidirectional relationship,this phenomenon was caused by emotional contagion. Furthermore,the statistical results showed that most of the unidirectional followees in Twitter were not acquaintances in real life but most of the bidirectional followees in Twitter were acquaintances in real life. The results of the Granger causality also suggested that the people,either acquaintances or not acquaintances,could spread their emotion to others in Twitter.