On Updates and Inconsistency Repairing in Knowledge Bases

The problem considered here is: given a consistent knowledge base and an update, i.e. some rst order formula, how can the facts in the knowledge base be changed such that (1) the update becomes true and (2) the constraints remain satissed. A fact modii-cation satisfying these two requirements is said to be a solution. We will present a technique to compute a solution for a given update and a given knowledge base. The salient features of our approach are: (i) we tackle the problem in a general way and also present a technique for repairing inconsistency, (ii) the user can interact with the system and actively innuence the solution to be generated, so we are not forced to generate all possible minimal solutions from which the user-nally draws the most preferred one, (iii) solutions are obtained by set-oriented fact processing rather than by single fact accesses to the old knowledge base, and (iv) in contrast to the other techniques proposed so far the consistency of the old knowledge base is exploited when generating a solution.