Research And Implementation Of Clustering Search Engine For Commerical Area

The statistics show there are above 1 billion products provided by e-commerce site on the internet.However the existing search engine for e-commerce site resent to the users a set of non-classified web pages. So it's difficult for users to find out what they need.Therefore,the search for products on e-commerce sites still bears a considerable potential for improvement.Because the method of web clustering can group the results into different categories automaticly,user can get a more valuable information through the categories.In this paper,we present and complete an intelligent cluster search engine for user to search products on e-commerce sites.The main achievements of the thesis are following:First, based on the analysis of the function requirement in the search engine,the architecture of the clustering search engine for e-commerce site is desigend.Then,study the relevant technology used in the system:MetaSeeker crawling workflow,Lucene retrieval workflow and web clustering method.Second, Analyzed the K-Means algorithm's advantage and shortcoming.And study affinity propagation algorithm, a new method in the processing of generating cluster set in affinity propagation algorithm is proposed based on the original AP. To study the performance of the proposed method,we did experiments and compared the three algorithm:K-Means,AP and newAP.The newAP algorithm obtains higher F-measure and lower Entropy, improves clustering execution time.So the proposed method is better than others.Third,when building a clustering model of this system, according to the text characteristics of the system, a suitable similarity model of the system was proposed.The last, complete the function of search engine:indexing,searching and clustering model on the foundation of lucene. The user interface is finished by PHP. Through the comparison of the designed search engine,the existing search engine and the clustering search enging,the experiment shows the designed search engine is more efficient.And after extensive testing for performance,the testing resutls prove the availability of the designed search engine.