Expert System for Learning Styles Diagnosis Using Dempster–Shafer and Bayesian Network

The lack of intelligent diagnostic tools for determining the learning styles of students at STMIK Multicom Bolaang Mongondow is the subject of this study. The absence of reliable, flexible diagnosis techniques makes it difficult for teachers to modify their lesson plans. To address this, a hybrid inference approach is put forth that combines Bayesian Networks (BN) and Dempster-Shafer Theory (DST) to manage uncertainty and offer tailored suggestions based on the VAK (Visual, Auditory, Kinesthetic) model. The main way the research contributes is by creating and confirming a new framework for reasoning that combines probabilistic and belief-based inference. System development follows the Expert System Development Life Cycle (ESDLC). The method's internal consistency and reliability are demonstrated through accuracy evaluations and white-box testing of the core inference mechanism. Stable performance across devices is demonstrated by functional testing. An 86.67% match rate is found through accuracy testing based on comparisons with expert evaluations. The findings support the usefulness of combining BN and DST to manage uncertainty and enhance adaptive learning suggestions. This method-focused contribution provides a useful framework for individualised learning systems in the future that don't rely on massive datasets