Research On Multi Agent-Based Information Retrieval Mechanism For Engineering Supervision

With the explosive growth of the network information, the precision of information retrieval is higher than before and the contradiction between the high precision and massive data becomes more and more obvious. In such circumstances, more and more research departments and scholars from home and abroad begin to study the mechanism of information intelligent retrieval. Although they have achieved some preliminary research results and even some successful experimental AI-searching, it is not enough. With the practical work, the thesis intends to set up the Industry-specific Intelligent Information Retrieval Model.For the intelligent auxiliary support of the diverse and dynamic information retrieval in the engineering supervision, the thesis develops a model named CE-BVSM (Classifier and Expand-Based Vector Space Model), which combines the theories of Agent and Multi-Agent with the ideas of Naive Bayesian Classifier, professional vocabulary, word vector control and the technologies such as vector space algorithm.The Model is designed by the B/S (Browser/Server) structure in the whole. By the means of the RDF and XML technology, it changes the past heterogeneous data to standardization and then establishes a metadatabase of information resources. Multi-Agent is the Model's intelligent aspect. All Agents will be divided by the function into main Search Agent, Site Agent, Data Collection Agent, Needs Analysis Agent, Pretreatment Agent, Monitoring Agent, Filters and User Agent, Communication Agent. The idea of weighted vector space algorithm is proposed, which is the innovation aspect of the Model. Instead of the traditional vector space retrieval method, which is simply the unicity of the keyword searching, the idea introduces such means as query expansion, in accordance with the keyword function size distribution of value, user interests weights and other means to retrieve information. Data coarse screening, information retrieval and information filtering and other multilayer information refining methods are used into the idea. The relevant technologies are also explained in the thesis. After the specialized intelligent retrieval mechanism finished, I establish intelligent information retrieval system for Heilongjiang Zhongtie Project Management Co. Ltd., provided strong guarantee for the further development of the company.