Recommendation system using hybrid fuzzy association rules for human smart cities

This work proposes a new approach for a recommendation system that uses a mixture of crisp and fuzzy sets for quantitative and qualitative features to improve recommendation suggests. In Business-to-Consumer (B2C) models, the identification of the classes of similarity between users is crucial for recommendation systems. In fact, recommendation systems use similarity between users to derive “guidelines” aimed at the suggestion of products. Unfortunately, when association rules are used for recommendation algorithms, a strong limitation descends from a set of “crisp”, for which a product can only belong to a group or not. Fuzzy sets can be used to overcome the limits of approaches commonly used in current recommendation systems, which provide a degree of belonging of a concept to a set, according to real values in a continuous range. One of the fields of application for this new approach is that of human smart cities that is a field of study and a concept for the implementation of city projects that began in Europe and spread worldwide over the past few years. It planning focuses on citizens’ wishes, interests and needs, not on technology alone. Technology comes later, after a clear definition of the benefits to the local citizens. It follows that knowing the preferences of citizens and their interests is fundamental to building human centered smart cities. Citizens continuously express their emotions and preferences on social networks. Extracting information from these databases in automatic and correlated form is an idea to dynamically trace the evolution of a human smart city.