RELATIONAL DATABASE ALGORITHMS AND THEIR OPTIMIZATION FOR GRAPH MINING
Data mining aims at discovering important and previously unknown patterns from datasets. Database mining performs mining directly on data stored in Data Base Management Systems. Several SQL-based approaches for (association rule) mining have been studied in the literature.
The main focus of this thesis is on the design and development of algorithms for graph mining (Subdue) using relational DBMS. We develop several approaches for discovering the repetitive substructures in a graph. Each approach is analyzed and optimized further to improve its performance. Two different approaches, cursor-based and User Defined approaches are studied in this thesis. The experiments evaluate these approaches and compare their performance with the main memory algorithm for a graph-based data mining (Subdue). The larger goal of this thesis is to achieve scalability.