Efficient service recommendation using ensemble learning in the internet of things (IoT)

Abstract

Internet of things (IoT) which has become widespread in various fields leads to a great variety in the provision of services to users. This variety in IoT services in turn pose challenges for users regarding their selections of appropriate services. Appropriate and proper service selection results in a better user satisfaction. In this regard, recommender systems (RS) are developed which help users to choose among various IoT services. Each user has his own characteristics in terms of profile and previous service selection, thus, customized methods should be proposed based on each user’s own characteristics and service interests. In this paper, for an IoT environment, which includes a large number of objects, services and users, some service RSs are proposed. In the proposed RSs, user profiles and previous users’ activities (in terms of service usage) are used based on which a binary decision tree has been made. The created decision tree is used for user classification in terms of their profile and service interest. The output of each RS is then ranked and combined using ensemble learning techniques, which improve the accuracy and efficiency of the recommendations. The result of ensemble learning is thus customized according to each user profile and service interest. The proposed RS is evaluated on a collected dataset and results confirm the high efficiency of the method presented in this research.