Graphs are becoming increasingly important to model many phenomena in a large class of domains such as chemo-informatics, bio-informatics, computer vision, communication network analysis and social network analysis. To deal with these needs, many mining approaches have been proposed for mining frequent subgraphs. These discovered frequent subgraphs can be used for various tasks such as characterization, classification and clustering of complex structures, building graph indices and performing similarity search in large graph databases. The prosperity of web2.0 and social media brings about many social networks of unprecedented scales. This presents new challenges for more-effective graph mining techniques. Many algorithms exist in literature for graph mining. Most of the existing algorithms perform well on medium size networks and fail on very large graphs. This is because of the enormous memory requirements due to the exponential number of frequent subgraphs possible. In this paper, we propose a frequent subgraph mining algorithm as an extension of the algorithm proposed by us[8]. It performs well on very large graphs. We evaluated the performance of our algorithm using synthetic data set. The experimental results show that our algorithm performs better than original version of the algorithm and also gSpan.
Volume 11 | 04-Special Issue
Pages: 1653-1660