异质网络社区发现研究进展*

异质网络将复杂系统中的信息抽象成不同类型的节点和链接关系,不同于同质网络,基于异质网络的社区发现能够挖掘出更加精确的社区结构。异质网络的社区发现通过对异质网络中的多维结构、多模信息、语义信息、链接关系等信息进行建模表示和提取分析,以发现其中相对紧密稳定的社区结构,对网络信息的获取与挖掘、信息推荐以及网络演化预测具有重要的研究价值。首先对社区发现当前研究的不足进行了简单阐述,接着引出了异质网络的定义；随后结合实例介绍了异质网络社区发现的现有研究方法,包括基于主题模型、基于排序和聚类相结合、基于数据重构和基于降维的方法等,并针对各类方法指出了其特点和局限性；最后讨论了当前该领域在结构复杂性、信息多样性、数据规模等方面面临的挑战。在将来,基于并行化、可扩展、动态增量的研究更能适应当前的变化环境。

Survey of community detection in heterogeneous networks

Most real systems consist of a large number of interacting, multi-typed components, while most contemporary researches model them as homogeneous networks without distinguishing different types of objects and links in the networks. Compared with the homogeneous networks, community detection based on the heterogeneous networks could obtain more accurate community structure. By modeling and analyzing various information including multi-dimensional structure, multi-mode information, richer structure and semantic meaning in heterogeneous networks, community detection is to detect useful and relatively stable community and valuable for network information collection and mining, information recommendations and predicting the evolution of networks. This paper introduced the community detection and the definition of the heterogeneous networks. In community detection in heterogeneous networks, the current mainstream methods included topic model, ranking-based clustering, data reconstruction, dimensionality reduction and so on. This paper summarized the above types of methods and analyzed their performance with practical applications. It also discussed the development trend of the community detection in heterogeneous networks. In order to adapt to the current environment, researches in the parallel, scalable and incremental dynamic heterogeneous networks will get more attention.