Recommendation Systems for Agri-Food Supply Chains: A Systematic Literature Review

Abstract

Agri-food supply chains face challenges in efficiency, sustainability, and resilience. Artificial Intelligence (AI), Internet of Things (IoT), and Blockchain-enabled Recommender Systems (RS) offer solutions. This systematic literature review uses the PRISMA protocol and eight research questions. It synthesizes the state of the art to identify fields of application, metrics, techniques, objectives, stakeholders, and critical gaps. The results show a dual application of RS. Internal applications focus on operational optimization, such as logistics, traceability, and waste reduction. External applications target personalization in marketing and consumer nutrition. Promising emerging techniques include reinforcement learning, graph models, and digital twins. However, scalable adoption is limited by gaps: lack of interoperability standards, a disconnect between academic validation and real-world contexts, issues in data quality and integration, and insufficient understanding of collaborative business models in multi-stakeholder ecosystems. The review concludes that future RS development in this sector requires a systemic, interdisciplinary approach. This should prioritize context-aware, user-centered design; holistic evaluation metrics that capture social and environmental value; and frameworks that support inclusive adoption—especially for small-scale producers—to build smarter, fairer, and more resilient agri-food chains.

Key words: Agri-Food Intelligence, Recommender System, Supply Chain.