Development of a methodology for optimizing elicited knowledge

1 1 ) !h15 piipcr a concrprual friiltlework and ;I)) ~)lxra~t;oiid ~~ir~l>tdology I S presen~ed FPI dcscrtbing ~ h c mosl .~pprol,i-i;ttc k ~ ~ o w l e d p rltcir.ir~on rnethod I ]~ r~? toco l , ir tervielv, ~nrlucrinu and repcrlvry g ~ ~ d ) for three c1:isxs or ~ k s (cl~:tgn~s~s, dcbuggirlg and 11-~~crpretation) a d fur exprrrs i r i l h ~t rc.n,c~ 115 n! various r'tct ors of cugnlil\e abilities. l!liing the cit.pcl~dcl~t \:nr iable: c l i . ( 1 ) total knowlzJgc ciiptured, (2) iin;c r t j ,icqitirr knoulrdgc. (3) knowledge qualily; (4) efficiency uf rt~e knoivledg c'l~c~~.~[ioii method: ;111d (51 ~mport;iace of tcsulring cf;tl;i, cxperin~et~~al results 111di~;ltr tlie various strcne:t~; r,I the four kiiot+lzdge tlicitatio~l ~nethods. The ki~o\vledp .~cquircd 15 also i~gnilicalltly atfectzd by the con~bincd I';ic!ors oC expcrr's strengths in dflercnt cogni~rvs Licto~s ' i ~ ~ d the nlrthocl ot-ktlowtedge cliciti~tro~l used B a d o n tt~r.,r liltd~np. a \l:~rchtny Indcx Lbr cornblning tasks. knuwlcc!g< :r:Itciirllit~i~ u ~ r l IloJ, i111d cuy~irive iibilitics ct tht' txpert is L ~ c s c I ; ~ ~ ~ . The r)u\conlt of \Ills research prov~des :heoretical aud p r :~~ \ i~a l ~rnplic;~tio~~s f o r FLulniln Computer I n~ r rac t~on (IICI) and ~rdlnitlg o f knowlrilge rngineers. 1. Ohj~ .c t ivcs and Significance Knowlctigc ellc~tation i s 111e first slop in building c ~ p c t t systems a n d i t is a ~na jo r bottleneck in tIlc construction of cxpcrt systelns (IIofflr~an 1987). .Altho~:gh many dfiercn! rncthods of k n ~ w l e d g z elicitation exist, the filllowing issuei arc not known aboiil thcse nlethuds: ( I ) which blowledge elic~tatlon lnctl~ods are morc s ~ ~ ~ t a b l c for dfferent tasks; ( 2 ) which Lnowledge e l i c i t ~ t ~ o n methods best cxlract ddTercnt kinds of knowlcdgc; and ( 3 ) how important of :I factor is the htrcrtgths of f a c ~ o r ~ in cognitive abilities of experts 111 rhc knowledge elicitat~on procedure. T h e p t i i~~ary ubjvcrivc. of [h is research I S to combine thc :bow thrcc qucs~inns Into n new cclnccptual rnodcl, N I I I L J I will provjdr n fraluework ,ind a r r l e t h o d o l o ~ for selecting the l?~ost appropriate knowledge clicitat~on tnethod for each t;~sk This derived conceptual rnodcl and ~ ~ i c ~ h o d o l v g ~ c a l Framrwark has an impwtnnt iinpsct on the d e s ~ y of human computer lntcraction dr,J training or knowlcdgc