Generalized Affinity-Based Association Rule Mining for Multimedia Database Queries

Abstract.

The recent progress in high-speed communication networks and large-capacity storage devices has led to a tremendous increase in the number of databases and the volume of data in them. This has created a need to discover structural equivalence relationships from the databases since queries tend to access information from structurally equivalent media objects residing in different databases. The more databases there are, the more query-processing performance improvement can be achieved when the structural equivalence relationships are automatically discovered. In response to such a demand, association rule mining has emerged and proven to be a highly successful technique for discovering knowledge from large databases. In this paper, we explore a generalized affinity-based association rule mining approach to discover the quasi-equivalence relationships from a network of databases. The algorithm is implemented and two empirical studies on real databases are conducted. The results show that the proposed generalized affinity-based association rule mining approach not only correctly exploits the set of quasi-equivalent media objects from the databases, but also outperforms the basic association rule mining approach in the discovery of the quasi-equivalent media object pairs.