Design And Implementation Context-Aware Web Service Recommendation

With the development of Web services, the massive Web services appear on the Internet. The kinds functions of services continue to increase,while providing the same or similar capabilities’services are constantly increasing,so that the usrs is in the situation of Web service information overload. When the user’requirement is not clear, recommendation is becoming a more and more important way to satisify user’s need. Personalized remmendation as the primary way to solve the information overload is also applied to the field of Web services. There are two types recommendations in traditional Web service recommendation process. The one is focus on QoS of service, ignoring the needs of the user’s interest. In contrast with that,the other considers the functional properties of the service but ignores quality attributes. In order to improve the recommendation effect, the system designed in this paper combines personalized recommendation and the context-aware technology, which is taking into account the user’s contextual information in the process, then ouptputs the optimal target recommended list of services under the user’s current context.In this paper, user behavior records, the corresponding historical context information, service’s functional and non-functional properties factors are considered to predict the target user’s Web service preferences in the current context. Firstly the relevant concepts of context-aware technology and the methods of combining it with traditional recommendation system are researched, then two context-based recommendation algorithms are proposed based on that. Since the traditional recommendation algorithm provides limited information in the context of the conditions, this paper introduces the context similarity to improve the traditional collaborative filtering algorithm. We get the filtered data by context similarity and improve the user similarity and score prediction formula to achieve context-aware recommendation based on collaborative filtering. In this paper, to solve the problem of Web service recommendation based on the single non-functional or functional property, a context-aware recommendation based on Web service properties is proposed to predict user contextual preferences by evaluating both functional and non-functional properties of Web service, which computes the similarity between the user’s contextual interest and service content and use’s preferences for services’non-functional property.Finally, this paper implements the prototype of context-aware Web service recommendation system based on two kinds of context recommendation algorithms which are proppsed in the paper. The Web service management module captures, analyses and storages the Web service information and context information. The context-aware module is responsible for system related context information processing. The two modules provide the data source for the service recommendation module and the recommendation engines implemented by the two algorithms score the candidate Web services. The recommendation module combines the results from different recommendation engines according to the strategy and returns the recommend services list to uers.