Design and implementation of fuzzy expert syste shell : an approach to support fuzzy matching and fuzzy comparison

Abstract

In this thesis we focus on the processing of fuzzy and/or uncertain knowledge and on the inference mixed with fuzzy rules/facts and crisp rules/ facts. Some measures are proposed for fuzzy matchings and fuzzy comparisons in processing of the mixed inference. We use the object-level fuzzy inference for propagating the fuzziness of attribute value and meta-level fuzzy inference for the uncertainty of rules and facts. An inference framework is also proposed which supports the object-level fuzzy inference and meta-level fuzzy inference with fuzzy matchings and comparisons. A fuzzy expert system shell is designed and implemented which uses the proposed measures and inference framework to process the fuzzy and uncertain knowledge. The expert system shell can handle the knowledge which is intermixed with fuzzy facts/rules and crisp facts/rules.