Backward Chaining Method in an Expert System for Cat Disease

Abstract

This study aims to design and develop a web-based expert system for diagnosing feline diseases using the backward chaining method—a goal-driven inference technique that traces symptoms from a disease hypothesis back to supporting facts. The system was developed using the Prototyping method, comprising six stages: requirements gathering, rapid prototype design, prototype construction, expert evaluation, final system development, and system testing. Knowledge was acquired through interviews with veterinarians and a literature review, resulting in a knowledge base covering four types of feline diseases and seventeen symptoms, represented as a rule base using IF-THEN logic. The system features three user roles—admin, veterinarian, and pet owner—with pet owners able to conduct consultations without logging in via a symptom-based Q&A interface. System validation involved Black-box testing to verify functionality and expert-led accuracy testing to validate diagnostic precision. Black-box testing results showed that all nine test scenarios performed as expected (100%), while expert accuracy testing demonstrated a high level of accuracy. Consequently, this backward chaining-based expert system serves as a reliable, efficient, accurate, and accessible tool for the preliminary diagnosis of feline diseases.