Research On Personalized Meta-search Basing On Clustering Interest Model

With the rapid development of web technology, a great deal of network resourceshave become available, search engine has been one of the import information servicestools in daily life. But, there are some problems: on the one hand, it can’t provide all theinformation that users need; on the other hand, the results has little concern with theusers’ interest. Meta-search Engine settled the problem of incomplete and incorrectsearch results. But through studying the current Modeling Technology of Meta-searchEngine, the user-interest model of meta-search engine is set up only based upon a singleuser, which causes the loss of common interests. It focuses on the user-query model,ignoring the preference of interest for the members of search engine, user training anduser-friendly social search users. For this reason, this article starts from the commoninterests of uses to work up a user clustering model designs and the implementation of aprototype system GMS (Group Meta Search) based on the clustering model.The main research works are as follows:1. Through analyzing the searching behavior, a method for calculation of thus interestsbased on the characteristics is given.2. Through extracting the main features of interests of multiple users, gives auser-group-interest model. It is an application of Beeferman algorithm to form thedifferent user groups, combined with the ontology model to create the user interestin the clustering model.3. Recommendation algorithm is given based upon the clustering model and theinterests of users. Combined with such information as preferences, collectionsand user-friendly of the members of the search engine users, supposes a sort ofMeta-search engine algorithms.4. Design and implementation of the clustering model of interest based on user searchengine GMS (Group Meta Search). Finally, in this paper, it uses different search topics to test on the GMS meta-searchengine. The experimental results show that the clustering model based on user interestin meta-search engine GMS can better reflect the common interest features. The resultof personalized service is also satisfied. Combined with trends in the demand of searchengines, the feature of future research will also be discussed.