Path: utzoo!attcan!uunet!cs.utexas.edu!csd4.milw.wisc.edu!uakari.primate.wisc.edu!ctrsol!sdsu!usc!ucla-cs!uci-ics!zardoz!tgate!ka3ovk!drilex!axiom!linus!mbunix!bwk From: bwk@mbunix.mitre.org (Barry W. Kort) Newsgroups: comp.lang.smalltalk Subject: Re: SmallTalk V/286 - Some questions.. Summary: We tried it. We like it. Keywords: Model-based and rule-based reasoning, continuous-valued logic Message-ID: <62754@linus.UUCP> Date: 6 Aug 89 13:19:33 GMT References: <378@DB0TUI11.BITNET> Sender: news@linus.UUCP Reply-To: bwk@mbunix (Barry Kort) Organization: The MITRE Corporation, Bedford, Mass. Lines: 27 In article <378@DB0TUI11.BITNET> MUHRTH@DB0TUI11.BITNET (Thomas Muhr) asks some questions about DigiTalk Smalltalk V/Mac: > Is the built-in prolog-engine of any use ? We use the Prolog classes for a rule-based diagnostic expert system which analyzes fault and overload conditions in a color-animated smalltalk model of a local area network (LAN). The basic LAN model is coded in Smalltalk. The user can instruct various components of the model to misbehave in interesting ways. The expert system then has to diagnose the problem based on observed behavior of the model as reported by Lanalyzer measurements of collision events and statistical counts of packets and acknowledgements. We are using continuous-valued logic, so the diagnostic expert system forms an evolving opinion about the hypothetical fault conditions as confirmatory and exculpatory evidence trickles in over time. At the present time, there are only a few dozen rules, and the expert system is of modest scope. But it demonstrates the potential for integrated model-based and rule-based paradigms for simulation, inference, and diagnostic reasoning. --Barry Kort