Research On Key Technologies For Intelligent Search Engine

As the sharp increase of network information resource, people fall across great difficult when searching needed information in the network. Search engine enhances the ability to locate and collect information, helps people find the information rapidly. However, existing information search engine have many shortages in search efficiency, information maintenance, information repetition and over loading. As a result, research on how to make good use of intelligence technology to supply user a better personalization search engine has become the urgent task of information resource construction. Based on studying the development and status of network search engine, and analyzing the shortage of the existing search engine, complete works as below: (1) Analyze the work principle and search strategy of real time search engine, introduce the famous dynamic search algorithm——"fish-search", then discuss the advantage of "fish-search" over normal depth-first algorithm , as well as the problem of search range randomization, which leads to search repetition and overlong search time. Based on the analyze of the flock behavior of fish school, define a parameter "dist" to control the search range and time, improve the standard "fish-search" algorithm, so the search process can adjust the search range adaptively. Simulative experiments show some improvements over the original fish-search algorithm in the search effect. (2) To implement the personalization recommendation for information retrieve, do research on forward neural network for instant corner classification. Based on strict mathematical analysis with the compute tendency of CC4, design a neural network for fuzzy corner classification (FCC) which can accept real vector as its input signals, where user interest model expresses as real vector. Through the FCC it can judge whether one user interest model belongs to multi class. (3)Put forward the strategy of scheduling the search engines in meta-search engine by genetic algorithm. Speeds up searching of genetic algorithm using niche genetic as reference, carry out the scheduling strategy in meta-search engine. Thus, the combinations of the actual search engines can be optimized dynamically and the whole performance of meta-search engine can be improved.