Multi-mobile Agent System-based Personalized Data Mining System

Following the explosion of information on internet and the complex of network envirnoment, the phenomenon appears that information is blasting, it is also indigent. Facing countless information, the customer can not find valuable information for themselves.How to find the potential and valuable information from a vast sum of information? How to acquire active and intellectualized special service? These are our business. The intergration of mobile agent, data mining and ontology is one of the best methods to solve these problems.As a domain model, ontology provides us a common comprehension about the domain concepts and the hiberarchy of concepts, also decrease our dependence on the comprehension technology of natural language.First, we introduce the concept about data mining, then analyze the algorithm of users classification based on clusters.After the introduction of concept of ontology, then analyzes the representation and building algorithm of user model based on ontology.Followed by the model of Multi-agents cooperating with each other based ontology, we introduce correlative technology of mobile agent.The last, we give the model of personalized data mining system based on Multi-Agent technology, and the corresponding algorithm. This model adopts the technology of Multi-Agent. First we classify the customer, then, build the special model for all kinds of customer. This system can classify customers that access this system, at the same time, it can surveil the customer's browse process and log the visited WEB and customer's feedback, analysis and modify special ontology of customer; according to domain ontology, it can provide a interactive interface that can help customer to study and build personal information requirement themselves. Many agent make up of a organic MMAS(multi-mobile agent system)by arrange and cooperation. By the way of distribute task, it can finish the task of distribute algorithm that can not be solved by traditional data mining system, increase efficicency of data mining and communication and decrease the burden of network on distribute computing.