Some methods for dynamically generating decision implications

Abstract

This paper focuses on generating decision implications and concept rules by approximate reasoning in a formal context. Firstly, the solutions of generalized modus ponens (GMP for short) and generalized modus tonens (GMT for short) based on full implication triple inference method are given, the MP approximate reasoning and MT approximate reasoning of decision implications based on GMP and GMT in decision context are proposed. An approximate reasoning method based on the attribute logic pseudo-metric is introduced, and the relation between pseudo-metric approximate reasoning and MP, MT approximate reasoning is discussed. Thirdly, the properties of MP and MT approximate reasoning of decision implications and concept rules are discussed, and it is proved that: (i) the conclusion of MP and MT approximate reasoning of decision implications are decision implications, and (ii) the conclusion of MP and MT approximate reasoning of concept rules are concept rules. Fourthly, some generation methods and algorithms of decision implications and concept rules based on MP and MT approximate reasoning are proposed. Finally, examples and experiments are given to show that the methods and algorithms are valid.