Qualitative simulators can produce common sense abstractions of complex behaviors given only partial knowledge about a system. One of the problems which limits the applicability of qualitative simulators is the intractable branching of successor states encountered with model of even modest size. Some branches may be unavoidable due to the complex nature of a system. Other branches may be accidental results of the model chosen. A common source of intractability is occurrence branching. Occurrence branching occurs when the state transitions of two variables are unordered with respect to each other. This paper extends the QSIM model to distinguish between interesting occurrence branching and uninteresting occurrence branching. A representation, algorithm, and simulator for efficiently handling uninteresting branching is presented.
Registration: ISBN 978-0-262-51091-2
Copyright: August 4-8, 1996, Portland, Oregon. Published by The AAAI Press, Menlo Park, California.