This is a Research Planning Grant. The research goal is to increase the domain of applicability of fuzzy reasoning by developing a methodology to advance the use of fuzzy reasoning in applications very different from control. Although a consensus has developed on the most useful fuzzy reasoning methods for fuzzy control, these methods do not have widespread applicability to general knowledge based systems. The success seen in fuzzy logic control can be extended to general knowledge base applications by tackling the limitations of the conventional fuzzy reasoning methods. Two of these limitations are the restriction that base rules must be compact for fuzzy reasoning to succeed and the focus on rule-based knowledge representation. It is intended that the first limitation will be eased by developing and testing a new method of fuzzy reasoning based on geometric principles, utilizing the proximity of the input and the rule antecedent to produce an output that has the same proximity with the consequent, and thus can operate in both compact and sparse rule bases. To deal with the second limitation, methods of shifting fuzzy reasoning from rule-based to object-based or frame-based knowledge representations will be explored. Based on the results of this work, further projects will be designed to realize the goal of extending fuzzy logic to general knowledge based systems.
