Approximate Knowledge-base/Database Consistency: M2-2~ An Active Database Approachl

Many AI applications populate their knowledge-bases with information retrieved from large. shared databases. This paper describes a new approach to maintaining consistency between objects in a dynamic, shared database and copies of those objects which are cached in an application knowledge-base. The approach relies on an intelligent interface to active databases that we call a Mediator for Approximate Consistency (MAC). The MAC has several unique features: (1) it permits applications to specify their consistency requirements declaratively. using a simple extension of a frame-based representation language. (2) it automatically generates the interfaces and. database objects necessary to enforce those consistency requirements, shielding the knowledge-base developer from the implementation details of consistency maintenance. and (3) it provides an explicit representation of consistency constraints in the database, .which allows them to be queried Ilnd reasoned about. The paper describes the knowledge-base/database consistepcy problem and previous approaches to dealing with it. It then describes our architecture for maintaining approximate knowledge-base/database consistency. including techniques for specifying, representing. and enforcing consistency constraints.