TOWARDS IMPLEMENTING VALUATION BASED SYSTEMS WITH RELATIONAL DATABASES

Shenoy's valuation based system provides a knowledge representation framework for a variety of applications. In this paper, we study the valuation based systems using the notions developed for relational databases. It is shown that the objects and operations in a valuation based system can in fact be conveniently expressed using relational database terminology. In this framework, the construction of a suitable Markov tree, for example, can be viewed as query optimization. This study explains some of the assumptions inherent in such a query processing, which can lead to an efficient implementation of valuation based systems.