中文摘要：

针对微博用户交互行为内容简单,传播速度快的特点,采用传统方法时由于用户活跃度与传递系数的关联性低,微博信息的检测目标不准确,导致微博用户交互行为确定不合理和无效的问题。提出基于交互行为的微博用户影响力改进算法。综合考虑了粉丝质量,以及关注、转发、评论等交互行为对用户影响力的影响。通过分析微博用户交互特征及其对用户影响力的作用,选取微博信息被接受率、用户活跃度、用户间交互强度、用户间影响强度及影响力传递系数等指标来度量交互行为对用户影响力的作用强度,进而依据PageRank算法原理构建新的改进算法。最后,利用腾讯微博用户数据,证明改进算法能较好反映用户转发、评论等交互行为,并能够削弱不活跃僵尸粉影响,更贴近人们对影响力的一般认知,因而更有效、合理。

英文摘要：

In view of the characteristics of simple content and fast propagation speed of the microblog user＇s interaction behavior, when using traditional method, the detection target of microblog message is not accurate due to the low correlation between user＇s degree of activity and the transfer coefficient, and causing the problem that the determination of microblog user,s interaction behavior is not reasonable and invalid. An improved algorithm of micro blog users＇ influence was proposed based on the interaction behavior. The influences of quality of the fans, as well as attention, forwarding, comments and other interactive behavior on user＇s impact were taken into account comprehensively. Through the analysis of the characteristics of microblog user＇s interaction and the function on user＇s impact, the accepted rate of the microblog messages, user＇s activity, user＇s interaction intensity, influence intensity between users and influence transfer coefficient, and other indicators were selected, to measure the effect strengthen of interactive behaviors on user＇s impact. Then, according to the principle of PageRank algorithm, a new improved algorithm was constructed. Finally, Tencent microblog user data were used to prove that the improved algorithm can better reflect the user＇s repost, comments, and other interactive behaviors, and can weaken the impact of non-active zombie fans, which is more closed to people＇s general perception on the impact, and thus more effective and reasonable.