Intelligent modifications in a distributed knowledge-representation architecture

Introduces a novel mechanism for the updating process in a distributed knowledge base to allow flexible responses to cooperative interaction. The paper focuses upon the intelligent change of a given problem description according to some interaction. This feature is important because one often does not have handy the complete problem specification right from the beginning. Another motivation is the necessity to cope with the detection of inconsistent situations (where no traditional solution is possible). Our approach realizes the philosophy just to change those parts which are really influenced directly by the modifications; so, an entire recomputation can often be suppressed. An advanced modelling of this feature is indispensable, due to the exponential complexity of global constraint satisfaction. We first recapitulate related work. We illustrate the constraint processing paradigm. We introduce the basic alterations to the constraint satisfaction algorithm which are proposed for intelligent updating. The actual procedure is depicted subsequently. Finally, we exemplify the reasoning technique.<>