Logic based knowledge processing systems such as PROLOG based expert systems have shown obvious drawbacks in performing conventional database tasks. Knowledge processing by deduction on a large set of given facts can be better performed by a deductive database system based on Horn logic and relational database theory. A concept is presented to extend an existing relational database system to make feasible the deduction of intensional data from a given extensional database. The deductive extension provides an extended view mechanism and the integration of integn'ly constraints and leads to an enhanced quety mechanism. Thus, the conventional database becomes more expressive, shows a higher degree of consistency, and is evaluated more efficiently.