微博网络中用户主题兴趣相关性及主题信息扩散研究

Research on Correlation of Users′ Topic Interests and Topic Information Diffusion in Microblog Networks

运用Twitter-LDA主题模型对新浪微博数据进行了主题分析,基于用户主题兴趣相关性的研究表明用户间的主题兴趣具有三度相关性,同一主题兴趣下三度以内粉丝的发文数随用户发文数增加而波动式增加,各度粉丝与用户主题兴趣的相似度随粉丝度数的增加而下降。通过分析比较不同主题类别微博的扩散差异,发现生活情感类的信息最受用户欢迎,不同主题类别微博被转发的概率存在显著差异,平均转发数相差可达10倍,微博信息扩散树中各类主题在微博信息扩散深度、扩散时间间隔和用户的扩散能力方面都表现出不同的特征。

The topic analysis on the Sina microblog data is studied by using the Twitter-LDA topic model. The analysis based on correlation of users＇ topic interests shows that topic interests between users follow the three degrees of correlation. Within the same topic interest when the number of microblogs that users publish increases, the number of microblogs that their fans within three degrees publish also increases in fluctuation, and the similarity of topic interests between users and their multi-degree fans decreases with the increase of degree. Through the analysis and comparison of the diffusion difference of diverse topic categories, we find that users prefer the information with lifestyle topic, reposting probability is significantly different among microblogs within different topic categories, and the average reposting count can be 10 times in difference. In microblog information diffusion trees, diffusion depth, diffusion time interval and users＇ diffusion ability all show different characteristics for microblogs with different topic categories.