Personalized Service Recommendation Centered On User Interest

The broad application of service-oriented architecture(SOA)and Microservices architecture(MSA)in service-oriented software engineering has fueled the rapid growth of web and cloud services and service-based systems(SBSs).Tremendous web and cloud services have been deployed all over the world.Finding the right web services becomes difficult and critical as the number of functionally-equivalent services with different quality values increases.Thus,service recommendation has become of paramount research and practical importance.Existing web service recommendation approaches employ Skyline technology?collaborative filtering technology and Content-based technology.Skyline-based recommendation technology mainly uses Skyline technology to select those services with the best global QoS value from a large number of candidate services and recommend them to users.Based on the historical behavior data of the user,the recommendation technology based on collaborative filtering identifies the neighbor user set for the target user and finally predicts the QoS value of services for the target user based on the set,and selects the Top-k service,recommend to users.Content-based recommendation techniques directly recommend those similar services for users based on their historical behavior data.However,those approaches have not addressed a critical and fundamental problem:In this paper,before running system,there is a problem that how to recommend services according to a system engineer's quality constraints,e.g.,response time,reliability,etc.Meanwhile,there is also a problem that most of research did not take into account.When we run system,if the quality of a service has changed,or a new service has been updated,how to inform the system engineer in order to improve the correctness of SBSs,it need to select the appropriate system engineer to recommend.In this paper,before running system,we first propose two basic personalized quality centric approaches for service recommendation,namely,KNN-based approach and DSL-based approach.The KNN-based approach employs the k-nearest neighbors technique to find web services with quality values most similar to the system engineer's quality constraints.The DSL-based approach employs the dynamic skyline technique to find representative services that are not dominated by any other web services according to the system engineers' quality constraints.To overcome the respective limitations of the two basic approaches,we propose two hybrid approaches,namely KNN-DSL and DSL-KNN,that combine the KNN-based and DSL-based approaches.Then be running system,we also propose two basic personalized quality centric approaches for service recommendation.That is KNN-based approach and RSL-based approach.The KNN-based approach also adopts the k-nearest neighbors' technique to find system engineer with quality constraints which is most similar to new service.The DSL-based approach adopts the reverse skyline technique to find system engineers for its quality constraints,the new service is not dominated by any other old candidate services,it can refer to the "representative" system engineers.To overcome the respective limitations of the two basic approaches,we propose two hybrid approaches,namely KNN-DSL and DSL-KNN,which combine the KNN-based and DSL-based approaches.Extensive experiments are conducted on a dataset that contains 2,507 real-world web services to demonstrate the effectiveness and efficiency of our approaches.