Tagging Resources, Tagging Communities

Abstract

Most of the existing social network systems require from their users an explicit statement of their friendship relations. In this paper we focus on implicit communities of Web users and present an approach to automatically detect such communities based on user's resource manipulations. This approach is dynamic as user groups appear and evolve along with users interests over time. Moreover, new resources are dynamically labeled according to who is manipulating them. Our proposal relies on the fuzzy K-mean clustering method and is assessed on large movie data sets.