An Overlapping Clustering Approach with Correlation Weight

Overlapping clustering works on the hypothesis that one object belongs to one or more clusters. It tolerates intersection among clusters and discovers overlapping information hidden in observed data as well. Most overlapping clustering methods dedicate to studying the strategy of discovering overlapping observations, and ignore the correlation of overlapping observation and different clusters. In this paper, an Overlapping Clustering approach with Correlation Weight (called OCCW) is proposed. Correlation weights are assigned to those clusters that one observation belongs to along with the multi-assignment procedure in our approach. Experiments on multi-label datasets, subsets of movie recommendation dataset and text dataset demonstrate that the proposed algorithm has a better performance compared with several existing approaches.