Identifying and Handling Structural Incompleteness for Validation of Probabilistic Knowledge-Bases

The PESI(I tllrobahilities, Expvrt Sysl.euls, I(uowledge, and htferent:e) system attempts t.¢, address sonic: of tl,e probh.ms in expert system oh.sign thr,,,gh the use of the Bayesian Knowledge, Base (Ill(B) representation. Knowlo:lge gathered fi’unt a domaio expert is placed into this framework am] iJd’ercncirtg t,nder uncertail,ty is performed ovt.,r it. ]lowever. by the l|al.ure of BKBs, not all knowledge is incorporaletl, i.e. the representation need not be a templet e representation of all combil,ations and possibilities of the knowh.dge, as this would be impractical in many real-world systems. Therefore. inherent in such a system is t|w problem of incomplet.o knowledge, or gaps withb, the knowledge I~aso where areas of lackis,g knowledge preclude or hind,.r arrival al. a solution. Some of this knowh:dgc is intentionally omitted because its not. nee&.d for inferera’lug, whib. othG,r knowh,dg~, is ~:rrono,)u.qy omitted I)ut n(,(’,,ssary for valid rc.sull.s. Intentional oniis~ion. a str,.,ngth of I}w BKJJ relm,Sonl.al.ion, allows tbr capl.uring only the n.h.val,t porliol,s t,f knuwlolge critical to ntodeling an exl~ert.’s knowh:dgv wit.hiti a domain. q’his research t)rol)OSo~, a method fi~r ha,ldliug t.hc htt t,.r lbrm of incompletetwss administered through a graphical inlerface. Tht’ goal i.-, t.o detv(’t il,cOml)h.tt.m..ss at,tl be correcl.ed b.,¢ ~.t kulowh.dgv ~.’ltgitleL’r ill art in|oil.ire fashion.