Systematic Literature Review on Recommender Systems in E-Commerce: Emerging Techniques, Popular Algorithms, and Key Challenges
Journal Title: Advance Knowledge for Executives - Year 2024, Vol 3, Issue 3
Abstract
Objective: This study examines advancements in recommender systems (RS) within e-commerce, focusing on emerging techniques, popular algorithms, and key challenges. Through a systematic literature review (SLR), 26 studies published between 2019 and 2024 were analyzed using PRISMA guidelines. Findings reveal that graph neural networks (GNNs) and hybrid models, integrating traditional and AI-driven methodologies, improve personalization and scalability. However, challenges like algorithmic bias and real-time computational demands persist. Future research should emphasize scalable algorithms and fairness-aware systems to ensure equitable recommendations and optimize economic impact. Methods: The review tracks the evolution of RS from traditional methods, such as collaborative filtering, to modern approaches incorporating deep learning, GNNs, and hybrid models. We advance the state of the art of recommendation systems in terms of recommendation accuracy, scalability, and personalization, and solve cold start problems and data sparsity challenges. The study follows PRISMA guidelines, and the relevant studies published between 2019 and 2024 were synthesized by purposive sampling of 26 articles. Performance metrics, including precision, recall, and F1-score, were employed to assesssystem effectiveness, with quality assessments emphasizing methodological rigor. Results: Deep learning and GNNs have great potential in improving RS capabilities, but their deployment in real time applications is limited by high computational requirements for accuracy improvement and sparse data handling. Conclusions and Recommendations: Combining matrix factorization and neural networks, hybrid models emerge as promising solutions. Nonetheless, scalability, algorithmic bias, and fairness remain significant barriers, necessitating optimizing trust-aware systems and bias mitigation techniques for large-scale environments. Future work includes scaling up hybrid models, incorporating fairness mechanisms to establish equity in recommendation results, and studying the economic effects of RS on the behaviour and commercial outcomes of users.
Authors and Affiliations
Panyavanich, K. , Chatnopakun, S. , Bhumpenpein, N. , Viriyapant, K. , Nuchitprasitchai, S. & Nilsiam, Y.
A Strategic Cooperative Model: The Case of Islamic Financing Hub
Objective: This case study aimed to propose a strategic cooperative model and discuss the challenges encountered by Belt Road Initiatives (BRI) and the need to address those gaps to unleash the potential and impacts of B...
Factors Influencing the Acceptance of ChatGPT Usage Among Higher Education Students in Bangkok, Thailand
Objective: Utilising perceived ease of use and perceived usefulness of the Technology Acceptance Model (TAM), considering facilitating conditions, attitude, as well as privacy and security factors, this study endeavours...
A Book Review: Principle of Economics (Asst. Prof. Tanpat Kraiwanit, Ph.D.)
Objective: This review aims to analyse the principles of economics from Principles of Economics, written by Assistant Professor Tanpat Kraiwanit, Ph.D., and explores related articles to each topic, such as critical think...
Systematic Literature Review on Recommender Systems in E-Commerce: Emerging Techniques, Popular Algorithms, and Key Challenges
Objective: This study examines advancements in recommender systems (RS) within e-commerce, focusing on emerging techniques, popular algorithms, and key challenges. Through a systematic literature review (SLR), 26 studi...
Standard Languages Teaching and Plurilingual Education: Implications for the Promotion of Multilingualism in Cameroon
Objective: Cameroon is a multilingual country with over 250 languages divided into four groups that determine the linguistic and cultural complexity of the country. In this context, official bilingualism has been preferr...