Research On Clustering Search Engine

With the development of computer and Internet technology, information on the Internet grows explosively, search engine has become the most important tool for information acquisition. However, the way of just listing result from the traditional search engine cannot meet the needs of customer, people have to search the long result list and be trapped into the information overflow crisis. Therefore, how to let the user more convenient and fast through the search engine to find information become a very popular needs.Clustering search engine provide the new idea to solve those problems. This search engine combine the similar data though clustering analyizing search result and mining the similar and different of data. Then search engine return the more rational result and give people more comfortable feeling.This paper focus on the design of clustering search engine, system structure and clustering anglorem based on the research of search engine and data mining. We desigen the structure of cluster search engine with combination of the rational solving plan and search engine technology. We realize the key function modules with the help of Carrot2 open-source clustering tools. In addition, we provide the function for download pages from the search engine which does not have public APIs, so the system is more universal.In order to solve the problem of efficiency and clustering effect, which is the most popular problem in search engine user group, this paper compare the result of two matrix decomposition method. From this experiment, this paper find out the factor affecting the search engine efficiency. And then with the comparation of STC based and MD based clustering algorithm on Cluster Contamination Measure, Topic Coverage and Snippet Overage,we checking the advantage of the MD based clustering algorithm.