A Recommendation System Based on Fuzzy Signature

In recent years, recommendation systems have become essential tools for managing the overwhelming volume of information users face daily. In the context of Technology Enhanced Learning (TEL), for instance, where learners can be faced with a large variety of resources, their ability to deliver accurate suggestions is crucial for enhancing user experience, learning engagement, motivations, and overall learning outcomes. This paper introduces a novel recommendation system based on Fuzzy Signatures. A Fuzzy Signature is a fuzzy relation representing user interests. A similarity metric, named user kindredness, is proposed to determine like-minded individuals who probably share same interests. The kindredness is used to establish user neighborhoods from which the recommendations are generated. For ease of comparison, the performance of the method is evaluated on the Movielens dataset and compared against state-of-the-art approaches, including collaborative filtering, probabilistic methods, and fuzzy genetic algorithms, demonstrating good improvements in terms of accuracy and error rates.