Integrating Knowledge Graphs into Case-Based Reasoning for Fertilizer Recommendation in Data-Scarce Agricultural Scenarios

Fertilizer recommendation in data-scarce agricultural contexts is a challenge to enhance both productivity and environmental sustainability. This study introduces a knowledge-enriched Case-Based Reasoning (CBR) system that integrates Knowledge Graphs (KGs) to formally represent agronomic concepts and infer semantic relationships. Unlike traditional CBR systems that rely solely on historical data, our approach improves case representation by combining expert knowledge, domain literature, and empirical data from 300 coffee farms surveyed in Cauca, Colombia. The methodology follows a systematic engineering approach that includes expert knowledge acquisition, semantic modeling using domain ontologies, and case enrichment. This integration enhances contextualization and reasoning in low-data scenarios. Our system constitutes a step toward the development of practical, explainable, and flexible tools to support decision-making in precision agriculture, and it is currently undergoing validation in collaboration with coffee growers in Colombia.