基于遗忘曲线的用户影响力时效性度量方法

User influence measurement method based on forgetting curve

社交网络中的用户影响力研究对于客观认识信息传播规律、分析网络成员和辅助舆情监控等具有重大意义。现有工作对影响力度量的时效性考虑不足,导致难以准确预测影响力的变化规律。针对此问题,提出一种基于遗忘曲线的用户影响力时效性度量方法,利用遗忘曲线和重复学习理论,并结合用户活跃度参数拟合影响力随时间下降及上升的变化过程。基于新浪微博和人人网的真实数据集,对该方法进行了仿真验证,结果表明:提出的方法能简便快速较准确地刻画用户影响力在一段时期内的变化情况。另外,虽然基于遗忘曲线的方法并没有实时准确预测出用户影响力的升高降低变化规律,但却能根据初始影响力和用户的活跃状态大致给出一个用户影响力的变化区间,有利于研究者对于用户影响力未来变化趋势作一个大致的预测。

User influence evaluation in social networks has great influence on learning information propagation law,analyzing users and monitoring public opinions. In the existing works, the timeliness of influence measure is insufficiently considered which leads inaccurate prediction for the change law of influence. This paper proposed a method to evaluate the user influence timeliness based on forgetting curve. Repetitive learning theory and user activity were used to fit the change curve of user influence along with time. Based on the real data set of Sina micro-blog and Renren, the simulation results indicate that the proposed method can depict the changing trend of user influence during a period in a convenient, quick and accurate way. Although the method based on forgetting curve does not predict the user influence, but it can give the trend of the user influence based on the range of active users.