TO FACILITATE KNOWLEDGE ACQUISITION

EXISTING METHODS FOR KNOWLEDGE ACQUISITION IN DOMAINS CHARACTERIZED BY PLANNING RATHER THAN BY CLASSIFICATION HAVE LARGELY RELIED ON ADHOC PROCEDURES APPLIED BY SKILLED KNOWLEDGE ENGINEERS RATHER THAN ON FORMAL METHODS. KNOWLEDGE ENGINEERING IN THESE DOMAINS HAS THUS BEEN SUBJECT TO SEVERE LIMITATIONS RELATIVE TO BETTER-DEFINED ENGINEERING PRACTICES: THE KNOWLEDGE BASE CANNOT BE VERIFIED BY THE EXPERT PRIOR TO THE CREATION OF A WORKING PROTOTYPE OF THE PLANNED SYSTEM, THE KNOWLEDGE ACQUISITION PROCESS CANNOT BE SUBJECTED TO AUDITING, AND THE FIRST RIGOROUS DESCRIPTION OF THE KNOWLEDGE BASE IS IN A HIGHLY IMPLEMENTATION-DEPENDENT FORM. THIS PROJECT WILL DEVELOP A FORMAL MODEL OF A PLANNING-TYPE KNOWLEDGE BASE THAT IS IMPLEMENTATIONINDEPENDENT, READABLE BY THE DOMAIN EXPERT AND THE AUDITOR AS WELL AS BY THE KNOWLEDGE ENGINEER, AND RIGOROUS ENOUGH TO BE TRANSLATED INTO AN ACTUAL IMPLEMENTATION USING ANY COMMERCIAL EXPERT-SYSTEM SHELL. THE INHERENT COMPLEXITY OF THE MANUFACTURING PLANNING PROBLEM MAKES THE CAPTURE OF THE ASSOCIATED KNOWLEDGE BASE AN IDEAL CANDIDATE FOR THIS INVESTIGATION.