Abstract:
Case-Based Reasoning (CBR) is used for the knowledge representation in the social and economic systems. The paper presents the literature review of case-based reasoning method and its application in various fields. The hybrid model of integrated case library and fuzzy logic inference is considered. The algorithm for generating fuzzy rules, where each linguistic input variable can take 3, 5 or 7 term-values, described by triangular membership functions, is examined. A new procedure for accumulating conclusions of conflicting rules obtained as a result of logical inference, is proposed. The classification accuracy of the proposed hybrid model has been studied for different data samples with different membership functions. The results of the experiments draw to the conclusion that the developed method of automatic training based on fuzzy output significantly increases the accuracy of the classification of precedents.
Keywords:Case-Based Reasoning (CBR), precedent, fuzzy logic, fuzzy set, fuzzy rules, inference, knowledge base, process of accumulation.