C2P: Clustering based on Closest Pairs
Αλέξανδρος Νανόπουλος, Yannis Theodoridis, Yannis Manolopoulos · 2001
In this paper we present C 2 P, a new clustering algorithm for large spatial databases, which exploits spatial access methods for the determination of closest pairs. Several extensions are presented for scalable clustering in large databases that contain clusters of various shapes and outliers. Due to its characteristics, the proposed algorithm attains the advantages of hierarchical clustering and graphtheoretic algorithms providing both efficiency and quality of clustering result. The superiority of C 2 P is verified both with analytical and experimental results. 1