User Navigation System Based On Active Service

Active Service is a kind of new technology proposed for the purpose of serving according to user's requirement. User personal service plays an important role in active service. This paper improved simple requirement analysis subsystem in traditional active service prototype system and proposed a personal user navigation system which can provide commended service for users actively by combing user interest model. The main contents are as follows:1) For the pretreatment of user's requirement, ICTCLAS (Institute of Computing Technology, Chinese Lexical Analysis System) based on multi-layer HMM (Hidden Market Model) for Chinese and English lexical analysis was adopted, in order to gain a keyword list. Vector Space Model was used to build text expression of keywords. In this way, vector comparability algorithm was employed to count the similarity between user's requirements and certain interest class.2) In this Navigation System, a user interest set was adopted to build dynamic model for every user. Added updated user interest and deleted interest which user is not interested in based on basic competition and threshold-controlled trigger rule in order to reflect the variety of user interests and track the changes of user interests.3) Aiming at the lack of traditional user classifying method, SVM multi-classifier by constructing DAGSVM extended BT tree was proposed, in order to classify users by user interests which have been stored in user interest file.4) Aiming at the advantages and disadvantages of different user classification methods, user attribute classification and user personality classification were combined to commend services to users. The classified mapping from user attribute classification to user personality classification was obtained via constructing the mapping matrix, which can implement dual classified coordination.The navigation system based on active service was proposed in this paper worked in the cooperation with different modules including user interest navigation, pretreatment of user's requirement, user management, classified statistic and user classification can well deal with user requirements, build user interest model and commend personal services and so on.