The Studying Of Agent Based User Interest Model

Today when the computer's software and hardware and network technology is developed rapidly, people's demand for the intelligence and individuation of software systems and information service has already become a trend. Under this situation, the computer is considered not only the human's "tool", but also should be human's assistant and "cooperative partner". Computer systems should stand at the user's position "intelligently", and according to user's actual condition, help user to analyze and solve his present problem. In information service field, "provide the user right information at right time and right place", which should be the supreme goal which the information technology researcher and developer pursue.To help user, we must understand user at first, including such individual characteristics as user's goal, knowledge, ability, interest, etc. Express these user's characteristics with appropriate form and data structure, which constitutes a user model. The course of acquiring, expressing and updating the information of user model, is called "user modeling". In this thesis, we use "intelligent agent" as the tool of user modeling. The autonomy and learning ability and communication ability of intelligent agent guarantee it can finish this task well.In this thesis, we have constructed a systematic prototype which base on intelligent agent and user model technology in the application field of "email automatically process". In this system, we use mail's characteristic vector, the classification of the mail and treatment tactics as the expression of user's interest characteristic. In the initial stage of system, user appoints the classification of every mail and system set up a user model according to the classification. When the user model is perfect, system will classify every new arrived email according to the user model, and report the result to user, accept theuser's feedback, then update user model according to the user's feedback.In this system, we have implemented three modules including user interface, data process and user modeling according to the theory of intelligent agent. The user interface agent is responsible for the task of interacting with user, receiving user's command and reporting the result to user; data processing agent is responsible for the task of network communication, managing mail and classifying mail; user-modeling agent takes the responsibility of modeling the user, maintaining and updating user model. We designed these agents according to the structure of "mixed structure agent", and implemented them with thread mechanism in Windows system. Because three agents need cooperation, we designed an "Agent Message Processor", as the platform of agent communication. Three agents cannot communicate with each other directly, but Message Processor hold every agent's information and own a "message reflection table", which can aid the Message Processor to know each agent can process what messages. Three agents and Message Processor formed a multi-agent system framework that is called "federation system".