Many 'state-of-the-art" expert computer systems function through the application of a finite set of deductive production rules to a static database of knowledge. For some well-studied fields such a system is sufficient, as there are no significant changes being made in the fundamental theory behind the knowledge represented in the database. An example of such a system is MYCIN, a medical diagnosis system. MYCIN"S database is made up of the associations between symptoms and diseases.¹ This system works because these associations are well-documented and supported by centuries of medical study.