A Novel Methodology for Clustering Using the Kullback-Leibler Index

High-dimensional data clustering is considered as a difficult task in data analysis. In fact, clustering aims to maximize main data retention while, seeking minimum space to present and display cluster. In this paper, Kullback–Leibler distance based on the information entropy is used and a novel methodology based on undirected weighted pages with high compatibility is presented. The proposed methodology was tested on three real data sets, indicating higher efficacy and accuracy in comparison with other algorithms. The results show that the methodology can be used for data clustering effectively, since indicated optimum operation in time complexity and clustering results.

کلیدواژه ها:

Information entropy ، Kullback–Leibler distance ، clustering ، social networks