A New Incremental Algorithm for Overlapped Clustering

摘要

In this paper, a new algorithm for incremental overlapped clustering, called Incremental Clustering by Strength Decision (ICSD), is introduced. ICSD obtains a set of dense and overlapped clusters using a new graph cover heuristic while reduces the amount of computation by maintaining incrementally the cluster structure. The experimental results show that our proposal outperforms other graph-based clustering algorithms considering quality measures and also show that ICSD achieves a better time performance than other incremental graph-based algorithms.