Towards a Software Architecture to Provide Hybrid Recommendations for Smart Campuses

Currently, several initiatives have been proposed in order to offer solutions for intelligent university campus environments. It can be said that smart university campuses are a subdomain of the smart cities domain, with some similar problems, but with specificities. Recommender systems identify and suggest relevant information to the user, recognizing their potential interests through specialized algorithms and presenting resources that align with these interests. In the context of intelligent university campuses, recommender systems have been used to define which systems and technologies should be implemented. From this scenario, the objective of this article is to present a software architecture, called SmartC, structured in different services, to provide the essential infrastructure for the application of several recommender systems and a variety of types of items. The services and layers of the architecture were defined especially for intelligent university campuses and divided into three distinct sections: the access environment, the recommendations management environment and the persistence layer. The recommendation algorithms integrated to this architecture are considered hybrids, since they incorporate two types of filtering: content-based filtering and collaborative filtering. When users request new recommendations, the type of filtering will be switched, ensuring that new features are suggested with each system call and avoiding throttling. The developed prototype was evaluated from real item data and showed significant accuracy in the recommendation process.