Research On Frequent Access Pattern Discovery Based On Hybrid Semantic Log Knowledge Base

The semantic web makes computers understand the content on the Internet, which improves the efficiency of Web usage mining effectively. It has become an important research field in artificial intelligence. As an important part of Web usage mining, frequent access patterns discovery can be used to dig out the user’s frequent visit behavior from a huge amount of Web usage data in different situations. And the mining results have great significance for the development of e-commerce, improving site management and enhance personalized service.This thesis is structured as followes, starting from the ontology and rules of semantic Web, focusing on the combination of it and the frequent Web access patterns discovery based on the hybrid semantic log knowledgebase, main work includes:(1) Improving the formal description of log Ontology.Event is adopted as the core concept to improve the formal description of log ontology, defining the ontology as a six tuple, and using the application rules to represent the domain relations in log ontology. The domain relation of log ontology is mainly defined by experts, not only can’t they guarantee content comprehensive but also meet the dynamics situation. Our improvement not only simplifies the contents of log ontology, but also accord with the demands of practical requirement.(2) Constructing semantic log knowledge based by the combination of log ontology and rules based on heterogeneous method.By using Datalog hybrid rules to represent the domain relations and user access behavior. and combine it with log ontology to construct hybrid semantic log knowledge base within the Datalog security constraints. This method overcome the weakness of ontology reasoning ability and dynamic semantic expression ability, and complements the advantages of them, which improves the expression ability and reasoning ability of knowledge base effectively.(3) Presenting the hybrid Datalog log ontology knowledge system, based on an approach for discovering the frequent Web access patterns is proposed.Proposing an approach for mining frequent Web access patterns based on the theory of ILP in hybrid semantic log ontology knowledge system. And through the core input events--refE, extending the web access pattern space to construct candidate pattern set, validating the pattern’s validity and calculate the support of pattern. Finally, discover the frequent Web access patterns.(4) The design and implementation of the Web frequent access pattern mining system based on hybrid semantic log knowledge system.Design and implement the frequent Web access pattern mining system by a Java programming language. The system includes the construction of hybrid semantic log knowledge base and frequent Web access pattern mining. Using ontology parser to generate the log ontology,rule parser and check rules security to generate application rules, to construct the hybrid semantic log knowledge base.Using the approach proposed previously to find frequent user access pattern from the knowledge base. Finally, through the experiment, verify the feasibility of the research.