Pervasive Computing Lab (PCL)

The need for internet and web services has increased notably in the last few decades. In many real-world scenarios where users are bombarded with choices, recommender systems play a significant role in assisting users in making better choices based on their preferences and interests. The applications of recommender systems include, but are not limited to music, book, movie, healthcare, travel, fashion, transportation, and shopping. Recommender systems mainly utilize users’ rating history to suggest items to the users. Nevertheless, various approaches have been studied and analyzed throughout the years to use other information as well as further enhance the suggestions. This paper provides a comprehensive study of recommender systems based on deep learning, in which the multiple approaches such as matrix factorization, graph neural networks, deep reinforcement learning, large language models, and transformers along with their problems and challenges are assessed.