Research Of Group Personalized Search Technology Based On Clustering For Open Access Resource

Along with the development of open access movement, open access graduallybecome an important way for the researchers obtaining academic information. Facingthe overladen results which come from the web servers, we should build up aneffective management and organization way, and remove the results records which arenot relevant with users’interests. As a result, users can locate their search goals.In this paper, based on the traditional open access resource servers’shortcomings,we propose a group personalized search server with clustering.Firstly, in order to solve the local optimum problem of partitional clusteringalgorithm, we propose a text clustering algorithm which bases on hidden topics. Basedon the k-means algorithm, we add some limiting conditions on the choice of clustercenters and build up a hidden topics framework. We realize the goal of combination ofcluster and classification, by mapping the cluster objects on the framework anddividing all the clustering objects into several classes according to the map values.Secondly, because the existent open access servers can not meet the demand ofpersonalized inquires, we propose a personalized inquires model. In the traditionalquery method, we add users’group information and the clustering objects’classinformation, and then we locate the users’query goal in their interested source classes.As a result, most data resource will not be retrievalled, which are not relative withusers’interests.Thirdly, using the users’profiles, we will filter all the return results and somerecords with high clicking ratio will be recommended to users who are in the samecluster. In order to be convenient for the users locating their goals, we show the resultsin a hierarchical clustering way.Lastly, we analyze and validate the proposed methods. In Myeclipse integrateddevelopment platform, we use java development tools to realize a personalized searchsystem which bases on the text clustering algorithm. By the experiments, we test thetheory of the proposed searching system, and analyze the system’s price cost. As a result, we find a direction for the further research.