SCALABLE AND EFFICIENT CLUSTER-BASED FRAMEWORK FOR MULTIDIMENSIONAL INDEXING

Aretty Narayana, M.Srinivasa Rao, R.V.Krishnaiah

Abstract: Indexing high dimensional data has its utility in many real world applications. Especially the information retrieval process is dramatically improved. The existing techniques could overcome the problem of “Curse of Dimensionality” of high dimensional data sets by using a technique known as Vector Approximation-File which resulted in sub-optimal performance. When compared with VAFile clustering results in more compact data set as it uses inter-dimensional correlations. However, pruning of unwanted clusters is important. The existing pruning techniques are based on bounding rectangles, bounding hyper spheres have problems in NN search. To overcome this problem Ramaswamy and Rose proposed an approach known as adaptive cluster distance bounding for high dimensional indexing which also includes an efficient spatial filtering. In this paper we implement this high-dimensional indexing approach. We built a prototype application to for proof of concept. Experimental results are encouraging and the prototype can be used in real time applications

Keywords: Clustering, high dimensional indexing, similarity measures, and multimedia databases

DOI: https://doi.org/10.15623/ijret.2013.0208028