Research Of Recommendation Approach For Service Mashup

In the areas of electronic commerce, City emergency contingency plans, scientific computing, etc., business users often need to integrate cross-domain data resources on the Internet rapidly according to their personalized demands. Unfortunately, such characteristics like distribution, autonomy and heterogeneity of data resources have brought lots of difficulties to the data integration. In recent years, the researches combine the service-oriented and the end-user programming and generate the new approach of data integration based on data service Mashup. It first takes the data service as the unified abstraction of various data resources to solve the heterogeneity of data resources and normalize their access methods. Then it realize data service Mashup to get data, process data and connect data with programming environment of data service Mashup. Though the data service Mashup tools usually are visual and graphic, it’s still challenging to solve the data integration problems for non-professional users due to the tedious process and the users’demand is not clear and so on. In this paper, we put forward a kind of effective method to recommend data services and Mashup pattern for users in the process of building data service, it will greatly alleviate the pressure of the users. The main research includes:1. Propose a dataflow-pattern-based recommendation approach for data service Mashup. It includes:define tagged-data service and data service Mashup plan and propose seven dataflow patterns based on data service Mashup plan we have defined; propose a recommendation approach based on dataflow patterns and actual requirement; make experiments to evaluate the effectiveness of our approach with the records in Yahoo!Pipes.2. Propose a connectivity-based recommendation approach for data service Mashup. It includes:propose three kind of relationship based on the data dependence and use them to describe the abstract data association of data services; propose a connectivity to assess all relationships of any data services; propose a recommendation approach based connectivity and actual requirement for data service Mashup; make experiments to evaluate the effectiveness of our approach with the records in Yahoo!Pipes and simulate some data to evaluate the efficiency of our recommendation method.Comparing to related work, the experimental evaluation demonstrates the utility of our method about service recommendation and it conforms to the users’ need. With the help of the recommendation method, users without IT experience can create data service Mashup conveniently and finish the data integration that users need smoothly.