Expert System for Student Talent and Interest Using Certainty Factor and Dempster-Shafer Methods

Abstract

Elementary education systems in Jepara Subdistrict currently lack standardized frameworks for identifying student capabilities, leaving educators and parents without reliable tools to recognize individual talents and interests. We developed a hybrid expert system that combines Certainty Factor and Dempster-Shafer methodologies to establish quantitative assessment protocols for elementary student aptitude evaluation. Our research employed a quantitative descriptive approach, gathering data through structured behavioral observations, educator interviews, validated questionnaires, and academic documentation from multiple elementary schools across the district. The system processes student behavioral patterns using Certainty Factor methods for initial inference, then applies Dempster-Shafer algorithms to combine evidence sources while managing assessment uncertainty and subjective evaluation parameters. Preliminary testing reveals the system can generate percentage-based aptitude measurements across various domains, with interest category evaluations reaching 37% in targeted areas. We evaluated performance through accuracy validation, expert correlation analysis, precision-recall calculations, response time measurement, and knowledge base quality assessment. The hybrid approach demonstrates measurable improvements in talent identification accuracy when compared to traditional subjective methods, establishing a quantitative foundation for evidence-based educational planning. The system offers schools a standardized capability assessment tool that reduces evaluation bias while optimizing resource allocation for personalized learning development. Educational institutions can implement the framework to support more objective decision-making in student guidance and curriculum planning, particularly valuable for Indonesia's evolving educational landscape that emphasizes individualized learning pathways