Abstract
Collaborative Filtering (CF) is fundamentally characterized by Recommender Systems (RSs), which have recently attracted researchers' attention. The ever-increasing data about users and items and the emergence of machine learning approaches have motivated the recent development of CF. The sparsity caused by the lack of recorded transactions and data makes it challenging for CF to distinguish between users' similar preferences. As a result of the data sparsity issue, CF ultimately lacks the ability to generate useful recommendations and suffers from poor performance. This paper proposes a novel model that uses clustering and artificial neural network to address the issue of data sparsity in CF. The proposed model CANNBCF, a short name for Clustering and Artificial Neural Network Based Collaborative Filtering, is evaluated using four different datasets from four popular domains (books, music, jokes, and movies). The proposed model shows its superiority to solve the sparsity issue that the traditional CF technique encounters. In this paper, eight experiments are conducted to evaluate the performance of CANNBCF. The evaluation criteria include accuracy, precision, recall, F1-score, and Receiver Operating Characteristics used to examine the proposed model. The results of the experiments show that CANNBCF effectively solves the sparsity issue, improves the quality of recommendations, and demonstrates promising prediction accuracy.