<![CDATA[Corn is a strategic food commodity in Indonesia, with production of 22.44 million tons in 2023. However, disease attacks can cause productivity declines of up to 30-80%, mainly from downy mildew, leaf rust, and leaf spot. The limited number of pathology experts in the field leads to delayed diagnosis, resulting in significant economic losses for farmers. This research aims to develop an expert system for diagnosing corn plant diseases using a hybrid Fuzzy Tsukamoto and Naive Bayes method to enhance diagnosis accuracy, taking into account uncertainty in symptom severity levels. The system was developed using Durkin's Expert System Development Life Cycle (ESDLC), which consists of six phases. A knowledge base was built from SINTA and Scopus-indexed literature, identifying five diseases and 17 symptoms. The fuzzy Tsukamoto method was employed for the fuzzification of symptom severity, utilizing three membership functions (intensity, coverage, and severity), after which Naive Bayes calculated the posterior probability. The hybrid score was calculated with 40% Fuzzy and 60% Bayes weights. The system was successfully developed with an interactive web interface. Accuracy testing using 30 validation cases yielded an accuracy of 86.67%, with 85% sensitivity and 88% specificity. Expert testing by three plant pathology experts gave excellent ratings (average 4.6/5.0) for diagnosis accuracy, knowledge base completeness, and usability aspects. The hybrid Fuzzy Tsukamoto and Naive Bayes method is effective for diagnosing corn plant diseases, achieving 86.67% accuracy, which is 6.67% higher than the Certainty Factor method and 11.67% higher than the single Naive Bayes method. This system can help farmers perform early diagnosis and reduce dependence on experts.]]>
Copyrights © 2025
