Deep Learning Based-Recommendation System: An Overview on Models, Datasets, Evaluation Metrics, and Future Trends

Abstract

The growth of data in recent years has motivated the emergence of deep learning in many Computer Sciences related fields including Recommender System (RS). Deep learning has emerged as the solution; overcoming the obstacles of traditional recommendation models. Deep learning is able to enhance recommendation quality by learning non-linear and non-trivial user-item relationship, and extracting deep and abstract feature representations for users and items. However, deep learning in RS is still new and flourishing. The contribution of this paper is two-folds. Firstly, we will be providing several insights on the advances of RS focusing on deep-learning models, datasets and evaluation metrics. Secondly, we expand on the current trend and provide several possible research directions in the field of deep learning-based RS.