Research On The User Model For Supporting Cross-System Personalization

Current development on Service-oriented Architecture, Peer to Peer and Grid computing technologies has raised new opportunity and challenge for the development of Web. It's especially important to provide personalization service to users in such an open and dynamitic service-oriented environment. The Web service technology can organize the resource in the web in loosely-coupled way. It provides an efficient way to enable personalization cross various systems also. But there are various service providers, and some of them have personalization function themselves. Currently, how to effectively implement cross-system personalization becomes a problem need to be resolved urgently. At the same time the new challenge has emerged in the area of user modeling, which is the core and foundation of personalization.The web service based architecture for cross-system personalization is introduced in this paper. Based on this architecture the user model for supporting cross-system personalization is researched deeply. Firstly, there are various service providers but no unified standard for user model in cross-system personalization environment. The structure and representation is researched aimed to these characters. To support reuse of existing user model in cross-system personalization system, a structure of multi-dimensional and self-adaptive user model is established and represented formally at three level, the cognitive level, computational level and domain level.Secondly, the management method of user model is researched. Since the user model in cross-system personalization system need communication with various user models in different service providers, the bidirectional transfer of information about user between computational level and domain level is researched. By transferring the features about user between computational leveland domain level according to directed weighted graphs, reuse of user model from different service provider is realized.Finally, aimed to the above characters of cross-system personalization, an update method is suggested to update the user model in the cross-system personalization system, according to the information of user gathered by local system and service provider which the user has interacted with.