Nonmonotonic reasoning has been a field of vigourous study for a quarter of a century with practical implementations emerging during the last 15 years. Improving the efficiency of these systems is an important step in them gaining acceptance beyond the research sphere. The use of lemmas - small results that can be reused later in a proof or derivation - might be one way of improving performance. We have extended an implementation of the Theorist nonmonotonic reasoning system with four sorts of lemmas: goods, nogoods, derived literals and potential crucial literals. In this paper, we report the results of experiments designed to test whether these lemmas provide general performance boosts. |
| Cite as: Hagen, R.A., Sattar, A. and Goodwin, S.D. (2003). Improving Search in a Hypothetical Reasoning System. In Proc. Twenty-Sixth Australasian Computer Science Conference (ACSC2003), Adelaide, Australia. CRPIT, 16. Oudshoorn, M. J., Ed. ACS. 45-53. |
 (from crpit.com)
(local if available)
|