H-rule mining in heterogeneous databases

Abstract

We examined the problem of applying association rules in a heterogeneous database. Due to heterogeneity, we have to generalize the notion of association rule to define a new heterogeneous association rule (h-rule) which denotes data association between various types of data in different subsystems of a heterogeneous and multimedia database, such as music pieces vs. photo pictures, etc. Boolean association rule and quantitative association rule are special cases of h-rule. H-rule integrates previously defined rule concepts and expands association rule mining from single dataset mining to database mining.