HIERARCHAL CLUSTERING AND SIMILARITY MEASURES ALONG WITH MULTI REPRESENTATION

A Lakshmi Deepthi, J.V.D Prasad

Abstract: All clustering methods have to assume some cluster relationship on the list of data objects that they really are applied on. GraphBased Document Clustering works with frequent senses rather than frequent keywords used in traditional text mining techniques.Similarity between a pair of objects can be defined either explicitly or implicitly. With this paper, we analyzed existing multi-viewpoint based similarity measure and two related clustering methods. The main difference between a traditional dissimilarity/similarity measure and ours could be that the former uses merely a single viewpoint, which is the origin, even though the latter utilizes many viewpoints, which you ll find are objects assumed to not have the very same cluster using the two objects being measured. Using multiple viewpoints, more informative assessment of similarity could well be achieved. Theoretical analysis and empirical study are conducted to back up this claim. Two criterion functions for document clustering are proposed dependent on this wonderful measure. We compare them several well-known clustering algorithms which use other popular similarity measures on various document collections confirming the good sides of our proposal

Keywords: Multiview Cluster, Document id, ClusterDistance

DOI: https://doi.org/10.15623/ijret.2013.0208012