Double layered genetic algorithm for document clustering

Abstract

Genetic algorithm for document clustering(GC) shows good performance. However the genetic algorithm has problem of performance degradation by premature convergence phenomenon(PCP). In this paper, we propose double layered genetic algorithm for document clustering(DLGC) to solve this problem. The clustering algorithms including DLGC are tested and compared on Reuter-21578 data collection. The results show that our DLGC has the best performance among traditional clustering algorithms(K-means, Group Average Clustering) and GC in various experiments. © 2011 Springer-Verlag.