Dynamic Inference And Recognition Of Knowledge And The Robustness Of Knowledge Expression System

Rough sets theory, as a new mathematical approach to the expression, study and induction for vague knowledge or vague data, has become one of the research focuses in the field of information sciences, for it has been widely and successfully applied to machine learning, knowledge discovery, data mining, decision support and analysis and so on. Practical system (such as financial system, risk investment system, medical diagnosis system, etc) problems are often dynamic, classical Pawlak rough set is static and helpless for dealing with dynamic problems, S-rough sets developed Pawlak rough sets, and provide a theoretical support for solving various dynamic problems (such as dynamic data mining, dynamic knowledge discover and system dynamic rough characteristic, etc) induced by element transfer or attribute transfer.Element or attribute transfer and element dynamic augment or reduction are two dynamic problems involved in knowledge expression system. Element transfer and attribute transfer are two of the primary concepts in S-rough sets theory, but almost all the current researches about S-rough sets theory were presented on the assumption that element transfer or attribute transfer has been mastered, the researches with regard to the reason and influence of element or attribute transfer are scarce until now. Meanwhile, knowledge expression system is a crucial part of various information systems, the aim of element dynamic augment or reduction is to construct a robust knowledge expression system, but until now, none satisfying constructing algorithm has been found. Based on the summarization of predecessor' works, this dissertation presented the researches primarily as follows:1. The reasons for element transfer and attribute transfer were analyzed; then by employing the propositional logic theory, the processes of element transferring and attribute transferring were analyzed when system is interfered dynamically by exterior interference factors; according to the interference effects, the exterior interference factors were classed to element interference and attribute interference; and then the logic inference model between element interferences and system elements, and the logic inference model between attribute interferences and system attributes were constructed. On the basis of the researches above, the logic inference model of knowledge and the logic inference model of S-rough sets were proposed. The results above can makes people to master the change of knowledge and to recognize the unknown knowledge or concept hiding in system dynamically, moreover to interference system positively and to make system approach the prospective change. The results above, as the emphasis of this dissertation, provide a theoretical support for system dynamic forecast, system dynamic decision and system dynamic control.2. The concepts of mature degree, rule significance and rule frequency ratio were proposed, and the constructing algorithm for robust knowledge expression system was presented on the basis of the rules constituted by the concepts above.