Automatic Classification Of Personalized Information Services In The Ontology And Multi-agent Technology,

Due to the rapid popularization of the Internet and WWW, people are encountering the problem of information explosion. Each one that uses the Internet would like to retrieve the desired information or receive messages in a more efficient manner. Therefore, personalization becomes a hot research topic in the field of e-server. In this thesis, we present a personalized Internet information classification system with the capability of automatic ontology construction.Ontology is popular to present information retrieval systems, but its construction is heavily dependent on the manual effort. To overcome this drawback, we propose an automatic construction approach for the job ofconstructing ontology. In such ontology based information classification scheme, a new term operation that combines the Tagger and WordNet is first used to analyze lexicon for extract meaningful feature. Then, a novel ideal of virtual Concept is proposed to organize the extracted features into specific concepts. Finally, the Domain-Concept hierarchy is constructed by employing the concept inheritance method to represent the ontology. According to the experimental results, the automatically constructed ontology can classify the Internet documents, for instances, e-mails and home pages, in a good accuracy.Another issue of this thesis is to develop a multi-agent based architecturefor Personalized information service. The multi-agent system consists of Interface agent, search & monitor agent, classifier agent, filter Agent and personal service agent to carry out the tasks. Each agent in the multi-agent system has its goal to achieve. The advantages of using multi-agent technology are shown below. Intelligent, Interface agent accept user's query request and display the query result, the interface agent another work is learning, recording and analyzing the user's interesting and behavior; Autonomy, Search & monitor agent retrieve the information form the famous search engine, such as Yahoo, lycos and so on. Moreover, the agent can notice when our interesting web pages changed; knowledge management, the system has a knowledge database to maintenance the domain knowledge, user knowledge and keywords information, and they can self-updated to adapt the user; Collaborative, the agents work together to achieve the goal.