The Research Of Multi-Agent Information Retrieval Model

With the rapid development and widespread of the Internet, the Internet provides us an uncountable number of documents which deal with information from a neverending list of topics. Thus to search such huge and quick change information on the Internet, more exact and faster retrieval tools should be developed. Search engine with crawler based on indexes vary in recall and is less precision. In order to overcome these problems meta search engines aiming at special requirements appear for information extraction, information filtering and integration of heterogenous information resources. But currently few search engines employ intelligent techniques to increase precision. With user modeling techniques more and more popular many personalized agent system for web browsing are currently developed. So it is a straightforward idea to incorporate the idea of user modeling with machine learning methods into web search services.First this thesis discusses the search enginee, analyses an abstract multi-Agent information-retrieval system according to machine learning methods based on user modeling. Then a multi-Agent Information Retrieval Model Based on Ontology which is implemented on JADE is provided. The model is described as an information space including a web site, search engine, agent platform, and agents. Different types of agents, such as interface agent, process agent, staticIRAgent, local search agent and search agent with mobile ability, coexist and interact by exchanging information and knowledge to find user relevant information. The model uses Ontologies to classify the domains of documents and to assist users to normalize their queries and put forword using co-occurrence theory to generate lightweight ontologies semi-automatically for information retrieval. At last the thesis gives the future works according to current research.