Research On Model Of Search Enginc Based On Multi-Agent

Along with the rapid development of Internet, the information resources in the web are becoming more and more abundant, and the Internet has already become a kind of necessary and important access for people to gain the information. However web information has some characteristics, such as vastness, distribution, complexity, which make user unable to completely understand huge and volatile information. In this case, the traditional search engines, whose functions are fixed, don't cast for the user's desire. User's expect to use a new generational search engine, which has intelligence and individuation.This paper has analysed the development and tendency in thefuture of the engine technology all over the world, joined together with Agent technique, which is the latest research field of AI technology, and brought out the search engine research based on intelligent Agent. Using many AI models, such as the interest model, multi-users cooperation, the author has realized the individual service in the search engine based on the framework mentioned in the paper. And the author introduces a sort of new search engine model which puts the "Push information" together. As are following:(1) analyze and study the characteristics and main framework of traditional search engine. Detailedly discuss the search engine developing tendency in the future, and then propose that the future search engine's tendency is a sort of individual, professional and intelligent search engine based on Agent.(2) Analyze and study the main kinds of agents and their characteristics. According to the character of search engine and the factors of realization, choose the Mulit-Agent model to realize the search engine system. And the experiment shows: Mulit-Agent model is very easily implemented, and it's the most practicallyAgent model, which also accords with the search engines self characteristics.(3) Based on the Agent model and search engine framework, improve some AI algorithms and ideas and propose web page characteristics withdrawing technique, which greatly improves the search engine intelligence and individuation of search engine. All the technique include:a)In the Robot, based on the characteristic of structure distribution of the web station and hyper link structure, emphasize the keyword and weight of the hyper link, and increase the Robots intelligence in searching web page and downloading it, and reduce the expense of the web search and download.b)Through the marked tree and the layering mark symbols, increase the system ability to judge the core meaning of web page.c)Adopting some intelligent algorithms, such as the users interest research, multi-users interest cooperation, increase the intelligence of user interface Agent and user's individuation, and bring the information "Push and Pull" idea into the search engine, and introduce a new search engine model.(4)considering the realization of the above algorithms and ideas of Agent, give a detail realizing algorithm and demonstrate it with diagram. Deepen the understanding and familiarity of Agent.