 This paper provides a systematic review of existing recommender system approaches that incorporate deep learning techniques, categorizes them into four types, introduces challenges and proposes a comprehensive framework for their research advancements, and discusses future developments in the field. This article was authored by Hong Nizhou, Fei Xiong, and Hong Xiuqin.