A Multilingual Rule-Based Web Decision Support System for Walnut Fertilization: Development and Scenario-Based Verification

Walnut (Juglans regia L.) fertilization requires the joint interpretation of orchard age, expected yield, soil fertility, leaf nutrient status, irrigation-water quality and soil texture. This study developed and verified Version 1.1.2 of a multilingual, rule-based web decision support system. A rule-based design was selected because the available source set provides diagnostic thresholds and management guidance but no field-calibrated training data, whereas explicit rules permit traceable calculations, boundary testing and cross-language equivalence checks. The macronutrient module combines author-defined age-specific base values, yield-related nutrient removal, leaf N, P and K factors, soil Olsen P and exchangeable K factors, and a heuristic age modifier. Analytical input requirements were standardized for soil-water measurements, Olsen P, neutral ammonium-acetate exchangeable bases and dry-matter leaf analysis. Five orchard scenarios generated 6.7-19.9 kg N da⁻¹, 5.1-13.9 kg P₂O₅ da⁻¹ and 13.0-43.5 kg K₂O da⁻¹. Automated Chromium testing comprised 105 assertions covering scenario-language equivalence, boundary behavior, rejection of non-positive pH and leaf values, soft plausibility warnings, stage-based scheduling, fertilizer conversion and localization; all tests passed. The exact base rates and correction factors remain author-defined prototype parameters, and sensitivity analysis showed that suppressing the age modifier would increase affected scenario outputs by 9.1-25.0%. The tool should therefore be treated as a transparent research prototype until independent linguistic review, regional calibration and field validation are completed.