DECIDE An Interoperative Environment for Distributed Diagnostic Expert Systems.pdfVIP

DECIDE: An Interop erative Environment for Distributed Diagnostic Exp ert Systems Daiki Kishimoto and Takahira Yamaguchi Scho ol of Information, Shizuoka University, 3-5-1 Johoku Hamamatsu, 432 JAPAN, Email: fkisimoto,yamaguti g@cs.inf.shizuoka.ac.jp ABSTRACT When one diagnostic exp ert system fails in solving problems alone, how can it get informa- tion available to improve itself from another similar diagnostic exp ert system ? This pap er presents the interop eration environment to supp ort the pro cess. First, inference primitive structure templates have b een presented to share inference structures b etween two diagnos- tic exp ert systems. Next, the co op eration metho d has b een presented, using the dierence arising in the context of the corresp ondence b etween inference primitives of an originator and those of a recipient. The wrapp er with conversion facilities has b een also provided, using a common domain ontology develop ed manually. After designing and implementing such an interop eration environment, the exp eriment has b een done b etween the following mo del-based like diagnostic exp ert systems in real task-domains, a trouble-sho oting exp ert system and an enterprise diagnosis (nancial management) exp ert system. The empirical results have shown us that the former supp orts the latter in nding a b etter solution. 1 Intro duction Many eorts have b een made to construct knowledge acquisition environments in order to build up exp ert systems successfully. Early knowledge acquisition environments, such as TEIRESIAS(R.Davis,76) and MORE(G.Kahn, S.Nowlan and J.McDermott ,85), tried to lo- calize bugs in pre-constructed rule bases by meta-level rules, thr