A Hybridized Graph Mining Approach

Abstract

Data mining analysis methods are increasingly being applied to data sets derived from science and engineering domains which represent various physical phenomena and objects. In many of data sets, a key requirement of effective analysis is the ability to capture the relational and geometric characteristics of the underlying entities and their relationships with vertices and edges, which provide a natural method to represent such data sets.In Apriori-based graph mining, to determine candidate sub graphs from a huge number of generated adjacency matrices, where the dominating factor is, the overall graph mining performance because it requires to perform many graph isomorphism test. The pattern-growth approach is more flexible for the expansion of an existing graph. © Springer-Verlag Berlin Heidelberg 2010.