DATA PRIVACY IN RECOMMENDATION SYSTEMS: AN ALGORITHM BASED ON DIFFERENTIAL PRIVACY
The area of recommender systems has multiplied in recent years. This increase and interest in this type of systems is justified by the possibility of recommending products, information, services or even people to certain Internet users, taking into account their characteristics and preferences. A correct recommendation of the contents makes all the difference to adapt the user to a certain system or even dissuade him from using it. In this way, recommender systems prove to be an important tool for the assertive involvement of a given user. However, for its correct functioning, it is necessary to observe the behavior of users, through the collection of information arising from their interaction with the system, a fact that may cause concerns in relation to this data and in relation to their privacy. In order for users to receive recommendations efficiently, personal information is provided to these systems, as this personal and private data may be at risk. Therefore, it is important that these systems provide protection mechanisms for these data and maintain their privacy, in order to guarantee security and privacy in the recommendations. Motivated by this problem, this work aims to propose a collaborative filtering algorithm model that provides for the preservation of privacy, based on the concept of differential privacy, in order to protect and maintain data privacy and maintaining the accuracy of the elements. recommended as well as the efficiency of the algorithm.