Study Of Agent-Based Intelligence Information Retrieval Architecture

With the development of Internet, people use a large number of information resources through Internet. However, with the rapidly increasing information in Internet, it is a challenge to select suitable information reformation from so many information resources. The current information retrieval technologys just offer keywords-based searching, but ignore the semantic content of the keywords itself. They still have many limits and their capabilities need to be enhanced. This paper proposes a Web intelligence information retrieval system on the basis of agent-WIIRS and describes the realizing method and architecture of- WIIRS.Agent technology is a new distributed computing technology, Agent is a computing unit which can run persistently and autonomously under distributed environments, it has several features such as initiative, interoperability, reactiveness and autonomy. Agent technology provides an effective approach to solve newly emerging distributed application problems, and in the mean while, it provides a reasonable conceptual model to study features of distributed computing system comprehensively and accurately. Therefore, Agent technology is the most effective solution to the information retrieval. For cooperation between agents to succeed, effective communication is required. KQML is one of common communication languages among agents. In this paper, we introduce agent communication mechanisms. By plugging XML into KQML , a more flexible and integrated representation of communication content is implemented, which can support the communication of KQML-based of agents. This method has higher flexibility andmore extensive application value.This paper proposes a Agent-based metasearch engine system in which multiple agents coordinate to attain information with intelligence. In the system, K-modes clustering algorithm is used to produce user's personalized information pattern and the personal pattern can be adjusted by user Agent automatically. This system adopts metasearch of structure of search engine, regards agent as the systematic basic module, utilizes autonomy and interaction of agent finish individualized Internet search of message of user. In system design, we present dispatcher tactics of member search engine that user fond of. These member search engines on the basis of mobile agent, which provide an environment for parallel processing, take full advantage of ontology which expands the requirement of users to the semantic words set and provide the document analyzer that can filter the Web pagers returned by the search mobile agent according to the certain algorithm. Consequently WIIRS presents the most relevant documents to the user. It overcomes confinements of traditional information inquiring system and can improve information retrieval performance significantly.Finally, this paper analyzes the significance of research on this system, and discusses the unsolved problems about this subject.