Heterogeneous ensemble learning method for personalized semantic web service recommendation

Abstract

Semantic web service (SWS) discovery and recommendation (SWSR) has emerged as a potential technology which aims to fulfill the user requirements by providing an improved recommendation for Academic and business communities. In SWSR, the user search pattern is adopted to make service discovery as well as recommendation. In order to achieve the precise recommendation of SWS, the ensemble learning method is utilized. This method encompasses the elimination of in-appropriate features and selects the optimal features of the requirements for the Academic and business communities. Semantic analysis is the one of the dominant technologies for SWSR, but it has not yet been explored by applying the ensemble learning over the service features to make optimal selection of features and to provide personalized recommendation. With this motive, in this paper a Heterogeneous Ensemble Learning method for Semantic web service Personalization and recommendation (HEL-SWSR) framework has been proposed. It will revitalize the industries to select optimal services SWS discovery. HEL-SWSR assists the feature extractions, concatenation of features selection using user profiles and triples from the OWL-S files. This framework combines various methods that eventually ensure service selection through the Maximum Voting Ensemble (MVE) technique. The MVE helps to select the services and recommends the top-10 services. From that list, the Academic or Business communities can be able to predict the appropriate services. The proposed framework performance is noticeably enhanced when compared with the traditional user search pattern technique.