A Greedy Strategy Link Clustering Method Based on AP

With the emergence of link clustering method (LC), research on link community detection has become an important topic in the overlapping community detection. After calculation of link similarity, LC method divides the community through hierarchical clustering and partition density method. However, it´s possible that the hierarchical clustering would fall into the local minimum, and that the segmentation density method will cause too small community division results. To solve those problems, in this paper, Affinity Propagation (AP) clustering method is used to replace the hierarchical clustering method. Then combined with the greedy expansion module strategy to improve the LC method. And then, a greedy strategy link clustering method based on AP(ALC) is proposed. Through comparing with LC method, CPM method and iLCD method with experiments on three real-world datasets, ALC is proved to have better performance on evaluations.