A Dempster-Shafer-Based Expert System for Personalized Higher Education Major Recommendation

Selecting a study program at a State University (Perguruan Tinggi Negeri/PTN) is a decision that many senior high school students approach without a systematic review of their interest, aptitude, achievement, and career goals, a gap made more pressing by the limited counseling capacity available at most schools. Earlier expert systems for major recommendation have relied on Forward Chaining or Certainty Factor, methods that combine evidence into a single rule outcome or scalar confidence without explicitly representing the uncertainty left by incomplete evidence. This study developed and implemented a web-based expert system, Peta Kampus, that applies Dempster-Shafer evidence theory to generate a personalized major recommendation from a student's answers to a structured questionnaire. The knowledge base was constructed with school counselors and an education practitioner into 275 evidence-linked questions covering 55 majors across 21 faculties, each carrying an expert-assigned belief value. The system was built on a native PHP Model-View-Controller architecture with MySQL as the data store, and its inference engine combines successive evidence using Dempster's rule of combination while tracking residual uncertainty and evidentiary conflict. The system was verified through Black Box Testing of its functional modules, with all 15 tested scenarios passing, and evaluated through an accuracy comparison against three independent evaluators (school principal and two guidance counselors, 90% average agreement across 30 cases) and a User Acceptance Test administered to 30 students, which returned an overall satisfaction score of 94.8%. The resulting system gives students and counselors a belief-and-plausibility-based recommendation that is more transparent than a single deterministic score.