基于关系和积网络的社交网络链接预测

Link Prediction of Social Network Based on the Multiplication Network of Relation Sums

针对传统的链接预测模型推理方法通常只针对简单网络,只考虑单一的网络结构特征,难以适应不断复杂化的社交网络的问题,提出一种基于关系和积网络的链接预测方法,利用和积网络快速精确推理的能力进行链接预测.实验结果表明:该方法可以真实的还原社交网络的用户关系,比传统统计关系学习方法方法具有更快的推理速度和更高链接预测准确率.

Traditional link prediction model can only be applied to simple network,which takes only single network structure characteristic into consideration,failing to deal with the increasingly complicated problems of social networks.In this paper,a method based on the multiplication network of relation sums is put forward to predict links by using the quick and precise reasoning ability of the multiplication network.The experimental results show that the method can restore the real relationships between social network users.Compared with the traditional method of statistical relation learning,this method has faster reasoning speed and higher prediction accuracy.