信息传播中的用户转发选择预测

Prediction of“Forwarding Whom”Behavior in Information Diffusion

在线社会关系网络中,用户之间的关注关系网络承载着上层的信息传播,关注关系网络的结构影响着消息的可见度,并影响着信息传播过程的转发选择。以新浪微博为例,围绕信息传播中的多次暴露现象展开研究,结合用户关注关系网络的结构,探索信息传播过程中多次暴露情形下用户转发选择行为的模式和规律。针对信息传播中用户在多个暴露源下的转发选择预测问题,融合消息内容、网络结构、时序和交互历史等多方面因素,建模和预测用户转发选择。实验结果表明,新方法的预测准确率高达91.3%。

On online social networks, follow- ship network among users underlies the diffusion dynamics of messages;meanwhile, the structure of underlying social network determines the visibility of messages and forwarding activitiesin the diffusion process. Taking SinaWeibo as an example, this paper focuses on multiple exposure phenomena ininformation diffusion, and investigates the patterns and regularities of users?? forwarding behavior among multipleexposures combined with the structure of follow-ship network. This paper analyzes the“forwarding whom”problemof users among multiple exposures in information diffusion, aiming to model and predict the forwarding behaviorof individuals, combining content features, network structure, temporal and historical information. The experimentalresults demonstrate that the new method achieves a high accuracy of 91.3%.