| With the rapid development of computer network and communication technology,mobile search has attracted much attention because of its convenience, accuracy,personalization and other prominent advantages. How to use mobile devices toretrieve useful information from vast amounts of information has become a veryimportant research topic. Due to the matching degree between the mobile user’s querywords and the document keywords is low, the result of the query is not ideal and thecontent is not targeted, that makes it difficult to meet the user’s needs. Therefore, theimprovement of mobile users personalized search accuracy has important theoreticalsignificance and practical value.However, to provide individualized informationservice for mobile users, the key lies in the understanding of users’ interests and theestablishment of user interest model.Firstly, this paper analysis related theory about personalized information serviceand agent. We use the User Agent as a modeling tool to mining the user interestpreference model through User-centric, based on area feature, with the aid of webpages based on topic preference vector differences and retrieve the nouns ofdifferentiation strategies effective method.Secondly, according to the characteristics of the mobile search, a mobile queryexpansion algorithm is proposed by utilize the mobile users around context. Thealgorithm based on LBS location service for users around information, computingcorrelation between entities and the query. In addition, on the basis of select highcorrelated words to expand the user query, we can effectively improve the mobileuser’s personalized retrieval accuracy rate and recall rate.Thirdly, we focus on personalized sorting algorithms, after added results feedbackmechanism and combined with the model for mobile user’s personalized searchalgorithm, the algorithm can give in the field of personalized retrieval outputaccording to the different user input, and update the user interest model according tothe related feedback mechanism.Finally, the huge system was divided into several autonomous Agent module afterwe have take advantage of the intelligent Agent the characters such as autonomy,self-learning and communication coordination, in order to design a prototype ofmobile personalization search engine. Furthermore, we developed those core modulesand effectively carried on the simulation experiment. Experimental results show that the degree of satisfaction can be increased by10%-25%. |
