Optimization of information bases.

摘要

A technique for increasing the efficiency of a system such as an artifical intelligence system or a data base system in which a query (Q) is applied to a set of formulas (URF). The technique creates an approximation (AF) of the set of formulas which is computationally more tractable than the original set. The original set is used to answer the query (627) only if the approximation fails to provide the answer (622). The approximation is made by a compilation process (613) which provides a series of ever-closer approximations. A given query is answered using the currently-closest approximation (623). The approximation may consist of a first set of formulas which is an upper bound (i.e. implied by) of the original set, and a second set which is a lower bound (i.e. implies) of the original set. In this case, the compilation process produces a series of increasing lower bounds and decreasing upper bounds until a greatest lower bound and a least upper bound are reached. Example applications of the technique are given for propositional languages, first order languages, lauguages based on definite clauses, and frame languages. <IMAGE>