A sampling-PSO-K-means algorithm for document clustering

Abstract

Clustering is grouping objects into clusters such that objects within the same cluster are similar and objects of different clusters are dissimilar. Several clustering algorithms have been proposed in the literature, and they are used in several areas: security, marketing, documentation, social networks etc. The K-means algorithm is one of the best clustering algorithms. It is very efficient but its performance is very sensitive to the initialization of clusters. Several solutions have been proposed to address this problem. In this paper we propose a hybrid algorithm for document web clustering. The proposed algorithm is based on K-means, PSO and Sampling algorithms. It is evaluated on four datasets and the results are compared to those obtained by the algorithms: K-means, PSO, Sampling+K-means, and PSO+K-means. The results show that the proposed algorithm generates the most compact clusters.