Building classification rules for case-based classifier using fuzzy sets and formal concept analysis

摘要: The focus of this paper is a construction better knowledge base in case-based classifier system. Our structure based on concept lattice where rules are built from its subconcept-superconcept relation. Since the can only be constructed inputs with binary attributes, descriptive and numeric attributes must transformed to attributes. In paper, we propose transformation using fuzzy set theory. We experiment benchmark data sets, Car Iris, determine performance term number used classification precision. results show that trend accuracy proportional size learning inputs. relatively small compared training data. produces very promising practice classify new problem more accurate than traditional classifiers.