Research And Implementation On Personalized Service Recommendation Mechanism Based On Social TAG

With the development of Internet technology, the number and type of services have increased rapidly. It has become a great challenge to help users find services with regard to their demands quickly and efficiently among a huge collection of services. Existing service recommendation approaches are mostly on the assumption of the acknowledgement of service sets that meet users’ demand for functionality, and use collaborative filtering techonology to recommend services with the best Qos value among services with similar or same functionality. On one hand, these methods ingore users’ personal preference for Qos. On the other hand, they can not recommend services to users to meet their functional requirements. Meanwhile, some newly released and high quality services that satisfy users’ demand are difficult to be recommened for the reason that they seldom be concerned or used.As a major application of Web2.0, social tag has an advantage of easy implementing, convenient to use and can dynamically adapt to the organization of network resources. It is helpful to evaluate services for users. Users’ tagging behavior on service can reflect their personal preference in some way and support the realization of personalized recommendation if we apply social tag to information services recommendation process. Therefore, for the shortcomings and insufficiency of the traditional service recommendation methods, this paper do the research on social tag based personalized service recommendation mechanism. The main contributions of the paper include the following aspects:(1) By analyzing the features of social tag, we introduce social tag to the filed of information service recommendation and build service-oriented social tag model to characterize the functional and non-functional attributes of services, which supports the further analysis of users’ personal preference and sevice features.(2) Based on the service-oriented social tag model, we analyze users’ personal service functionality preference and service functionality features to propose a service functionality-oriented personalized recommendation method. The method discovers target user’s neighbor users by considering users’ use frequency for service at first. Then it clusters services that target user and his neighbor users have ever used according to service functionality vector. Finally the method calculates target user’s preference for service classes by using users’ behavior on service functionality tag and recommend service sets with regard to target user’ personal preference. Moreover, cluster method improves the opportunities of recommending new services.(3) In the premise of recommending service sets that meet users’ demand for functionality, we consider users’ personal requirements for Qos and the dynamic of Qos to further propose a Qos-oriented personalized recommendation method. The method clusters services with similar or same funcitonality into several Qos classes according to their Qos features and finds neighbor users by analyzing users’ usage on Qos classes. On one hand, it can help to alleviate the data sparsity problem in the process of discovering neighbor users. On the other hand, it is helpful to recommend services to target user with regard to its personal requirements for Qos based on neighbor users’ usage behavior on services.In summary, this paper has deeply focused on the personalized service recommendation problem. We build the service-oriented social tags model and propose the functionality-oriented personalized service recommendation method and Qos-oriented personalized service recommendation method. The simulation experiments and service recommendation system implementation have shown the availability and effectiveness of the method in the paper.