Q-analysis clustering

Abstract

This article presents a novel cluster­ing algorithm, based upon topology concepts and Q-analysis. The algo­rithm allows to create a structural clustering, which does not depend on the knowledge of similarity mea­sures, distances, etc. The clustering is directed in a natural fashion by the internal topology of the network/ graph. Such topology becomes evident by applying Q-analysis, by means of the structure vector, which shows the network’s internal Q-connectivity.