中文摘要：

针对利用二部图刻画在线社会网络用户节点之间的多维度关系特征时的局限性,提出利用超图的数学理论建立用户行为的超网络模型。通过实体用户、用户活动、用户兴趣三维度的映射关系,更真实地刻画了在线社会网络用户行为,并提出利用个体用户兴趣度表征共同兴趣内容的方法挖掘出用户的潜在朋友关系。某论坛的真实数据验证了超网络模型能够快速定位用户并刻画出用户兴趣爱好差异性。因此,该模型对在线社会网络热点话题的规模以及用户隐性关系的挖掘和预测提供了依据。

英文摘要：

For the limitation of bipartite graph in depicting the muhidimensional relationship feature between the nodes of online social network users, we propose to build a super-network model using mathematical theory of hypergraph. It more really depicts the behaviours of online social network user through the mapping relationship of three dimensions ： the entity user; the users behaviour and the users interest. Moreover, we also propose to mine potential friend relationships of users with the method of using individual users＇ interest to represent the common interest content. The actual data from a certain network forum verifies that the super-network model can rapidly locate users and depict difference between users＇ interests. Therefore the model provides the basis for the size of hot topics in online social networks and the mining and prediction of users＇ covert relations.