Imprecise Database Inference Using Functional Dependencies

Knowledge discovery in databases can be enhanced by augmenting them with \catalytic relations" conveying external common sense knowledge. Catalytic inference analysis ? the systematic analysis of inference closures in databases augmented with catalytic information ? uncovers new facts and rules, and latent inference channels. This paper presents a formalism for analyzing imprecise inference based on functional dependencies (FDs) in \mixed" databases containing precise, imprecise, and fuzzy information. The inference formalism is exible and robust, and has major applications in database mining and database security control. The ability to deal with mixed databases is especially signiicant. Imprecise or fuzzy information may be introduced to catalyze inference in precise databases.