Geo-Expert Plant: An Adaptive Expert System Based on Geolocation for Diagnosing Tropical Plant Diseases in Indonesia

Abstract

Plant diseases are one of the main factors contributing to the decline in agricultural productivity in Indonesia, directly affecting national food security. In tropical regions like Indonesia, the spread of plant diseases is strongly influenced by geographic factors, microclimates, and local cropping patterns. Early detection and accurate diagnosis are essential to prevent the wider spread of diseases, but the limited number of agricultural extension officers and access to experts often cause the diagnostic process to be slow and less accurate. This study introduces Geo-Expert Plant, an adaptive expert system based on geolocation designed to provide contextual plant disease diagnosis. The system integrates a forward chaining inference method with spatial filtering based on the user's GPS location. Through this approach, the system filters the most relevant potential diseases based on the characteristics of each specific region, allowing for more precise recommendations. The case study was conducted on three main crops—rice, chili, and tomato—across five major agricultural zones in North Sumatra: Medan, Deli Serdang, Berastagi, Simalungun, and Padang Lawas. The system's knowledge base was built from 100 real cases verified by field experts and local agricultural authorities. System testing showed an accuracy rate of 92.4% compared to diagnoses by human experts. By leveraging a combination of expert knowledge and spatial factors, Geo-Expert Plant contributes meaningfully to the digital transformation of the agricultural sector. The system is contextual, adaptive to local conditions, and easily accessible to farmers through digital devices. It is expected that this application can serve as a smart technology solution supporting precision agriculture and food security in the digital era.