A Fuzzy Signature-Based Approach for Recommendation Systems

In today's information-rich era, recommendation systems are essential for identifying and suggesting items tailored to user preferences. This paper introduces a recommendation approach based on Fuzzy User Signatures, enabling the development of effective recommender systems, even in scenarios where user data are limited. Fuzzy User Signatures are used to compactly represent user interests and preferences and to calculate similarity between users. The practicality and effectiveness of this fuzzy signature-based approach are showcased through its application in a movie recommendation system case study. An experimental evaluation using the MovieLens dataset demonstrates a significant improvement in performance compared to a similar state-of-the-art probability-based approach.