Is knowledge refinement different from theory revision

摘要: In addition, we are grappling with the issue of whether these two sub-areas have dierent goals. We believe that the goal of KBR is the renement of potentially noisy Expert System KBs; whereas TR systems rene more complex KBs, which may involve recursion (making programs themselves renable), but which are inherently noise-free. The main part of this presentation will consist of a discussion of the (general) characteristics of the two types of systems, under a series of headings where their strengths and weaknesses are noted.Representation Schema Used: Whereas TR systems have used both propositional and FOPL representations, KBR systems have used the propositional calculus and production rule representations. The TR community have claimed this as a great plus; on the other hand, most real-world applications of ML have needed only the power of the propositional calculus.