Computing and Informatics

Context-Aware Recommender Systems (CARS) are extensions of traditional recommender systems that use information about the context of the user to improve the recommendation accuracy. Whatever the specific algorithm exploited by the CARS, it can provide high-quality recommendations only after having modeled the user and context aspects. Despite the importance of the data models in CARS, nowadays there is a lack of models and tools to support the modeling and management of the data when developing a new CARS, leaving designers, developers and researchers the work of creating their own models, which can be a hard and time-consuming labor, and often resulting in overspecialized or incomplete models. In this paper, we describe GUMCARS -- a General User Model for Context-Aware Recommender Systems, where the main goal is to help designers and researchers when creating a CARS by providing an extensive set of User, Context and Item aspects that covers the information needed by different recommendation domains. To validate GUMCARS, two experiments are performed; first, the completeness and generality of the model are evaluated showing encouraging results as the proposal was able to support most of the information loaded from real-world datasets. Then the structural correctness of the model is assessed, the obtained results strongly suggest that the model is correctly constructed according to Object-Oriented design paradigm.