Abstract: There are many relationships among users of social networks. These relationships determine the division of user??s community structure in a network. To detect community structure in multiple relationships social networks accurately, a community detection algorithm in multiple relationships social networks based on the study of information propagation process in multi-subnet composited complex network model is proposed. According to information propagation in the multiple relationships social network, the nodes in the network can be transformed into vector which can be suitable for clustering algorithm, and then clustering algorithm can be used to detect community structure in multiple relationships social networks. The algorithm takes the interaction of multiple relationships in the network and the interaction between heterogeneous nodes into consideration, and generates the information matrix which can represent the influence of each node on the entire network. The nodes with similar influence would be divided into the same community structure finally. The experiment results show that, compared with the traditional community structure detection algorithm, the proposed algorithm can not only improve the accuracy, but also divide heterogeneous nodes into a community to mine hidden information in multi-relationship social network according to different needs of users.

Key words: complex network, community detection, information propagation, multiple relationships online social network, multi-subnet composited complex network

摘要： 社交网络的节点之间存在着多种关系，这些关系共同决定了网络中节点的社团结构划分。为了准确地发现多关系社交网络中的社团结构，通过研究信息在多子网复合复杂网络模型上的传播过程，提出了一种多关系网络中的社团结构发现算法。该算法基于多子网复合复杂网络模型建立的多关系社交网络，利用信息在多关系社交网络中的传播过程，将网络中的节点转化成能够被聚类算法处理的向量形式，进而采用聚类算法完成多关系社交网络中的社团结构划分。该算法综合考虑了网络中多种关系的相互作用以及异质节点间的相互影响，得到的传播信息量矩阵表示了各节点在整个网络中的影响力，并将影响力相似的节点划分到同一个社团结构中。实验结果显示，与传统社团结构发现算法相比，该算法不仅在准确度上有所提高，还能将异质节点划分到一个社团中，可以根据用户不同需求挖掘出多关系社交网络中的隐藏信息。

关键词: 复杂网络, 社团结构发现, 信息传播, 多关系社交网络, 多子网复合复杂网络模型