Sistem Pakar Diagnosa Penyakit dan Hama Tanaman Terong Berbasis Website untuk Petani Pemula

Abstract

Novice farmers often have difficulty identifying diseases and pests of eggplant because of limited knowledge and restricted access to agricultural experts, which delays treatment and increases the risk of crop failure. This research aims to design and develop a web-based expert system for diagnosing eggplant diseases and pests using the Dempster-Shafer method to handle symptom uncertainty. Data were collected through observation, expert interviews, and literature review, resulting in 10 types of diseases and pests, 25 symptoms, rules, and belief values for each symptom. The system was developed using the Waterfall method with PHP and MySQL and modeled using the Unified Modeling Language (UML). In the diagnosis process, users select the observed symptoms and the system combines their belief values using Dempster’s rule of combination to display the disease or pest with the highest confidence level and its treatment. In the calculation example with three symptoms, Bacterial Wilt obtained the highest belief value of 0.924. Black Box Testing achieved a 100% success rate from 28 test scenarios, while White Box Testing produced cyclomatic complexity values between 1 and 10 across 11 modules. A questionnaire completed by 13 respondents gave a feasibility score of 90%, indicating that the system is feasible.