Application Research On Meta Search Engine Based On Agent

It is well known, the Web information increases exponentially with the rapid development of the Internet, and there are many ways of getting information from the Internet. How to retrieve information correctly and effectively is widely concerned by researchers and users. The existing independent search engines have the defects of low database coverage and low recall ratio. The meta search engine expands the scope of retrieval by scheduling several independent search engines, and improves the recall ratio to some extent. However, the independent search engines return to a huge number of repeated and redundant results, so that the result display agent of meta search engine bears more burden. Besides, the precision is lower, and the response time is longer in the system.In order to solve the defects, agent technology is introduced and applied to meta search engine, and an intelligent meta search engine system model combined agent is established in the paper. The new meta search engine system makes full use of the agent, intelligent and autonomy attributes, and combines the agent advantages in information mining and retrieval, to improve the query performance and efficiency of meta search engine.The main research contains as follows:1. Introduce the knowledge of Agent and MAS, and the concept and principle of meta search engine etc. Besides, the paper analysises and summaries the research present situation about agent technology in the meta search engine application field at home and abroad, and puts forward the problems of traditional meta search engine exists.2. Establish an intelligent meta search engine system model based on reward mechanism. Member agents are created for each member search engine.The member agent collects the results of member search engines and makes corresponding process.3. Put forward a kind of intelligent member search engines scheduling strategy adaped to the system model:the member search engines scheduling strategy based on reward mechanism. The scheduling strategy fully considers the factors affecting the query performance of meta search engine, ranks the query important degree of each member search engine in accordance with certain reward mechanism, and selects the several member search engines with higher important degree for scheduling. 4. Put forward a kind of intelligent results merging strategy adaped to the system model:the retrieval results merging strategy based on reward mechanism. The results merging strategy is mainly aimed at the retrieval results returned by member search engines scheduled, and merges and sorts the retrieval results in accordance with the comprehensive matching degree rank between search results and request.5. Realize the collaboration and communication among Agents with KQML language in the system, and ananlysis and test the system performance, and prove that the intelligent meta search engine based on reward mechanism is better than traditional search engines in the aspects of recall, precision and response time to a certain extent.