Deep learning for Context-Aware Recommender Systems: A Systematic Literature Review

Abstract

Due to innovative advancements and an increasing amount of digital information over internet, recommendation systems (RS) are gaining more attention to find relevant information in a short instance. Currently, context-aware recommendation systems (CARS) are making a significant contribution to item recommendations based on user context. The addition of context in RS is a challenging task that not only broadens its dimensionality but also creates more sparsity. Recently, the field of deep learning (DL) has gained more focus due to its revolutionary improved efficiency in various domains. Nowadays, due to the efficient functionality of DL, Rs used it to produce high-quality recommendations. The DL based CARS facilitate improved ways to capture user preferences, items and user interaction histories and item features. In this systematic literature review (SLR), a review of studies relevant to deep learning-based context-aware recommender systems is presented in a systematic manner.  The current work synthesises the recent studies that cover the advancement in the field of context-aware recommender systems based on diverse variants of deep learning techniques. The SLR also discusses the application domains where DL based CARS perform significant contribution. The SLR concluded by discussion, based on research questions that highlighted twenty-two DL techniques which can be integrated in CARS to improve recommendation accuracy. The results have discovered nine domains where DL is used in CARS to provide higher efficiency. Finally, the study presented some future suggestions on the basis of current scenarios of DL utilization in CARS.