Recent trends in recommender systems: a survey

Abstract

In an era where the number of choices is overwhelming on the internet, it is crucial to filter, prioritize and deliver relevant information to a user. A recommender system addresses this issue by recommending items that users might like from many available items. Nowadays, the prevalence of providing personalized content to users through a website has increased profoundly. The majority of such websites use recommendation models to reduce a user’s searching time. Many new recommendation models are being proposed to address the changing business requirements of eCommerce organizations. Recommender systems can be broadly classified into three categories, i.e., clustering-based, matrix-factorization-based, and deep learning-based models. Many scopes and use cases are available where recommendation models play a vital role. The advent of graph representation learning and LLMs hinders recommendation models from being more effective in promptly providing relevant suggestions. This survey comprehensively discusses various deep learning-based recommendation models available for different domains. We also discuss the pros and cons of popular recommendation models. We also discuss various open issues of recommender systems and outline a few future directions. This study also provides insight to explore novel and helpful research problems related to recommendation systems. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.