Abstract:

Although collaborative filtering has been used widely in recommender systems, it suffers from several problems such as high dimensionality in the data ratings, vulnerability to attack, and sparsity. To improve the performance of collaborative recommenders, Formal Concept Analysis (FCA) has been applied.  However, the traditional FCA-based methods require high cost of run-time as they are based on strings comparison included in the recommendation. This paper proposes a new FCA-based method for collaborative recommendations based on prime number. The use of prime numbers improves the comparison process since the comparison between numbers is faster as compared to string comparison.