A neighborhood-based clustering by means of the triangle inequality

Abstract

Grouping data into meaningful clusters is an important task of both artificial intelligence and data mining. An important group of clustering algorithms are density based ones that require calculation of a neighborhood of a given data point. The bottleneck for such algorithms are high dimensional data. In this paper, we propose a new TI-k-Neighborhood-Index algorithm that calculates k-neighborhoods for all points in a given data set by means the triangle inequality. We prove experimentally that the NBC (Neighborhood Based Clustering) clustering algorithm supported by our index outperforms NBC supported by such known spatial indices as VA-file and R-tree both in the case of low and high dimensional data. © 2010 Springer-Verlag Berlin Heidelberg.