Propagating Probabilities in System P
Rachel A. Bourne, Simon Parsons · 1998
In this paper we suggest a way of using the rules of System P to propagate lower bounds on conditional probabilities. Using a knowledge base of default rules which are considered to be constraints on a probability distribution, the result of applying the rules of P gives us new constraints that were implicit in the knowledge base and their associated lower bounds. Introduction Default reasoning has been widely studied in artificial intelligence, and a number of formalisms have been proposed as a means of capturing such reasoning (Ginsberg 1987), most prominent among which are default logic (Reiter 1980) and circumscription (McCarthy 1980). Many of these systems, including default logic and circumscription, have proposed particular mechanisms for default reasoning, and might therefore be considered quite specialised. However, there has also been work on more general approaches which attempt to analyse in broader terms what default reasoning involves. An early attempt to do this was Sho...