Research On The Knowledge Discovery In Conceptual Hierarchy Knowledge Base Based On The Multiple-Level Association Rules

Data Mining(DM), also called Knowledge Discovery in Database(KDD), is a new research domain which has appeared with the development of database and artificial intelligence technology. To discover association rules from database is one of the important methods of data mining.Comparing with the typical method of discovering association rules, the research has been developed from discovery of single-level association rules to that of multiple-level association rules. It is possible to discover new abstract rules by implementing the concepts of more abstract levels. In many applications, this method can be used to discover rules related to different levels of database.To combine knowledge discovery with knowledge acquisition is a new effective method for solving the bottleneck problem of knowledge acquisition for expert system. According to the concept of multiple-level association rule discovery in multiple-level knowledge database,this paper uses a high-level knowledge discovery method as an assistant method of knowledge acquisition for expert system. Then the system design, the concept-hierarchy structure encoding, and algorithm optimization have been implemented. The achievements of this paper include:1) On the basis of concept hierarchy, we propose a method for data preprocessing and domain knowledge representation, that is, the hierarchy structure encoding method. Then we put forward a new concept for knowledge discovery in knowledge base which views domain knowledge represented by concept hierarchy as discovering targets.2) We propose a new idea which considers the discovery in knowledge base as a method of knowledge acquisition in an expert system, that is, a new method that adds those discovered rules directly into the expert system's knowledge base in the form of production rules. Then we give a main frame on which we can apply knowledge discovery to the expert system.3) Combining the theory of concept hierarchy with the discovery algorithm of multiple-level association rules, we design and realize a KDCHK system based on multiple-level association rules discovery. The system is capable of the concept hierarchy knowledge representation for various domain-knowledge and of making transaction table. And fundamental capability of knowledge discovery system has...