中文摘要：

针对在线社会网络中用户间的关系存在多种关系复合的情况，应用复杂网络理论设计了结合协同过滤推荐算法和基于网络结构推荐算法的混合推荐算法。利用复合多子网模型和在线社会网络演化模型设计的推荐系统用户模型，并在此基础上提出了基于在线社会网络社团结构的最近邻查询算法，并可根据最近邻集合评分后做出推荐。仿真实验证明，该算法具有较高的准确率和召回率。

英文摘要：

In the online social network, there are many kinds of relationships. In this paper, a hybrid algorithm is introduced to design the complex network. This algorithm combines the two algorithms, one is a collaborative filtering recommendation algo- rithm, and other is based on the characteristics of the network structure recommendation algorithm. Based on the above resear- ches, a recommendation algorithm for multi-relationship online social network is proposed. User model of the recommendation system is derived from multi-relationship online social network evolution model. Based on the user model, the nearest neighbors query method of community in multi-relationship online social network is proposed, recommendations are made based on the nearest neighbor set and items set selected by the nearest neighbors. Experiments prove that evaluation criteria of recommended system, such as, recall rate, precision rate and so on are higher than those of traditional collaborative filtering recommendation system, and its prediction is more accurate.