Community Detection Analysis of Heterogeneous Network

With the rapid development of information society, intricate relationship between objects establish huge heterogeneous networks. The linkage is affected by multiple factors, which makes community detection on heterogeneous network a difficult task. Traditional clustering algorithms focus on divided factors, ignoring the combination of them. If the structure of multi-dimensional information is taken into consideration, the results can be more accurate and meaningful. In this paper, we introduce an improved fuzzy clustering algorithm to solve the problem of community detection of heterogeneous network. First extract the features of heterogeneous network and initialize K clusters. Then use a model to create a K-dimensional vector for each object which denotes the probability of belonging to every cluster. Through modifying a classic fuzzy clustering algorithm FCM (Fuzzy c-means) called HFCM, objects can be reassigned to cluster based on the maximum probability. Finally synthetic data and real data are used to verify the correctness of the algorithm.