Hard and Fuzzy c-Medoids for Asymmetric Networks

Medoid clustering frequently gives better results than those of the K-means clustering in the sense that a unique object is the representative element of a cluster. Moreover the method of medoids can be applied to nonmetric cases such as weighted graphs that arise in analyzing SNS(Social Networking Service) networks. A general problem in clustering is that asymmetric measures of similarity or dissimilarity are difficult to handle, while relations are asymmetric, e.g., in SNS user groups. In this paper we consider hard and fuzzy c-medoids for asymmetric graphs in which a cluster has two different centers with outgoing directions and incoming directions. This method is applied to a small illustrative example and real data of a Twitter user network.