Research And Implementation On Results Clustering Optimization Of Meta Search Engine

Nowadays, with the development of application service in network, information retrieval has become one of the main purposes of netizens use Internet service. Because the traditional search engine has some shortcomings and its own limitations, restricting user to obtain the resources. To improve the coverage of user search information and the accuracy of the search results, people start to pay special attention to Meta Search Engine based on search engine. But now most of them present the search results to the end user with linear list, as there are thousands of the search results, it make user spend much time to find what they really want. The main reason is that the search results aren't classified and reframed according to user's query custom and retrieval experience. Accordingly, clustering the search results is one efficient solution to improve the lookup speed and fast locate the required information.The thesis designs and implements a Meta Search Engine system with clustering called CMES(The Clustering Meta Search Engine)combining with association rules and clustering analysis by researching on Meta Search Engine technique and data mining technique, and it illustrates the specific implementation in detail.Building up search engine parameter database realizes the transform from retrieval requests to target search orders and the automatic extract of related contents, then the system extracts the main keyword sets after segmenting the subject and abstract of the search results to build up a AWM(Associated Word Matrix)and express the result feature vector based on boolean function and TFIDF function, then testing and analyzing the two vectors under the different clustering arithmetics. Basing on k-means and FCM arithmetic, the thesis presents k-means and FCM optimizing methods of results clustering based on the AWM(k-means_AWM and FCM_AWM), then testing and analyzing them. Afterward it chooses better one to clustering the search results. It gives some practical examples and analyzes the performance of the system. Finally, it summarizes the present study and suggests the system should be further improved, also it gives a direction for further study.