Reasoning About Assumptions in Graphs of Models

Solving design and analysis problems in physical worlds requires the representatio n of large amounts of knowledge. Recently, there has been much interest in explicitly making assumptions to decompose this knowledge into smaller Models. A crucial aspect of problem-solving paradigms based on such models is that they include methods to automatically, and efficiently, change models when the initial choice is found to b e in error. We represent physical domains as Graphs of Models, where models are th e nodes of the graph and the edges are the assumptions that have to be changed i n going from one model to the other . This paper describes the methods used in the Graphs of Models paradigm for changing models . The methods are based on the facts that errors contain information on how parameter values are to be changed and tha t assumption changes contain information on how they affect parameter values . This knowledge can be represented qualitatively, permitting fast inferenEe mechanisms th...

• Type: Conference
• Year: 1989
• Where: IJCAI
• Authors: Sanjaya Addanki, Roberto Cremonini, J. Scott Penberthy
• Keywords: IJCAI 1989, IJCAI 2007, Models Paradigm, Parameter Values, Smaller Models