UNCERTAINTY HANDLING AND COGNITIVE BIASES IN KNOWLEDGE ENGINEERING

GENERALLY DEFINED, KNOWLEDGE ENGINEERING IS THE PROCESS OF ELICITING KNOWLEDGE FROM HUMAN EXPERTS AND ENCODING THAT KNOWLEDGE IN MACHINE USABLE FORM. ALTHOUGH THE TERM KNOWLEDGE ENGINEERING CAME FROM THE ARTIFICIAL INTELLIGENCE (AI) LITERATURE, THE GENERAL PROCESS OF ELICITING AND ENCODING EXPERT KNOWLEDGE IS NOT UNIQUE TO AI. VIRTUALLY ANY INFORMATION SYSTEMS ENGINEERING EFFORT WILL, AT SOME POINT OR ANOTHER, REQUIRE EXPERT INTERVIEWS AND THE ENCODINGOF EXPERT PROCEDURES. DESPITE ITS IMPORTANCE, RELATIVELY LITTLE IS KNOWN ABOUT PROPER PROCEDURES FOR KNOWLEDGE ENGINEERING. MOST OF WHAT IS PUBLISHED ARE RECOMMENDATIONS BASED ON INDIVIDUAL CASE STUDIES AND OTHER ANECDOTAL EVIDENCE, AND NOT ON SYSTEMATIC INVESTIGATION. IN THIS RESEARCH, THE RESEARCHERS EXPLORE THE EXTENT TO WHICH COGNITIVE BIASES HAVE AN IMPACT ON KNOWLEDGE ENGINEERING, PARTICULARLY IN THE AREA OF UNCERTAINTY HANDLING. SPECIFICALLY, THEY WILL DEMONSTRATE THAT NOT ONLY ARE EXPERTS SUBJECT TO BIASES WHILE SOLVING PROBLEMS, BUT THAT THEY ARE ALSO LIKELY TO EXHIBIT "REPORTING BIASES" WHEN THEYARTICULATE THE KNOWLEDGE THEY USE TO SOLVE PROBLEMS. IN ADDITION, KNOWLEDGE ENGINEERS WILL BE SHOWN TO BE SUBJECT TOTHE SAME BIASES AS DOMAIN EXPERTS AND THEREFORE UNLIKELY TO CATCH BIASES IN KNOWLEDGE ELICITED FROM DOMAIN EXPERTS. AN AUTOMATED KNOWLEDGE ELICITATION TOOL (AKET) WILL BE DEVELOPED AND USED IN THE RESEARCH EFFORT.