Implementing Integrity Constraints in an Existing Belief Revision System

SNePS is a mature knowledge representation, reasoning, and acting system that has long contained a belief revision subsystem, called SNeBR. SNeBR is triggered when an explicit contradiction is introduced into the SNePS belief space, either because of a user’s new assertion, or because of a user’s query. SNeBR then makes the user decide what belief to remove from the belief space in order to restore consistency, although it provides information to help the user in making that decision. We have recently added automatic belief revision to SNeBR, by which, under certain circumstances, SNeBR decides by itself which belief to remove, and then informs the user of the decision and its consequences. We have used the well-known belief revision integrity constraints as a guide in designing automatic belief revision, taking into account, however, that SNePS’s belief space is not deductively closed, and that it would be infeasible to form the deductive closure in order to decide what belief to remove. This paper briefly describes SNeBR both before and after this revision, discusses how we adapted the integrity constraints for this purpose, and gives an example of the new SNeBR in action.