Design of a Mobile Application with a Hybrid Recommendation System for Connecting Andean Farmers and Urban Consumers in Peru

Mayta J.; Boronda A.; Montes J. · International Journal of Advanced Computer Science and Applications (2026) · DOI: 10.14569/ijacsa.2026.0170974

Resumen

The dependence on intermediaries within agricultural marketing chains not only reduces producers' profits but also generates consumer distrust regarding the quality and traceability of the products they obtain. This work proposes a mobile application with a hybrid recommendation system aimed at connecting farmers and consumers in Peru. The combination of collaborative filtering and content-based filtering achieves personalization of the shopping experience based on the producer's offer and the user's interests. Likewise, mobile geolocation technology is incorporated to support delivery tracking, together with an offline-first approach to support the application's operation in rural areas with limited connectivity. A preliminary evaluation of the recommendation system, using the Amazon Reviews 2023 dataset (100,000 interactions), compares the hybrid model against collaborative filtering, content-based filtering, and a popularity baseline across RMSE, F1-score@5, catalog coverage, and diversity. The results show that collaborative filtering provides the strongest performance among the models offering meaningful catalog coverage, while the fixed-weight hybrid configuration evaluated in this study did not outperform it, identifying weight optimization as a priority for future work. The offline-first architecture maintains a 100% synchronization rate even under intermittent connectivity conditions, providing preliminary technical evidence under controlled conditions. © (2026), (Science and Information Organization). All Rights Reserved.

Instituciones: Universidad Peruana de Ciencias Aplicadas