APE
A Computational Framework Integrating Psychometric Student Modeling, Educational Ontologies, and Explainable Bayesian Updating
EdTech / AI-Driven Pedagogy
Overview
This research-driven project develops an adaptive recommendation engine for personalized education. It constructs psychometric models of individual learners-capturing cognitive abilities, knowledge gaps, and learning styles-using item response theory and latent trait modeling. These student models are mapped onto a formal educational ontology with explainable Bayesian updating for dynamic instructional recommendations.
Key Features
Psychometric Student Modeling
Latent trait and cognitive diagnostic models for proficiency inference from assessment data.
Educational Ontology Integration
Machine-readable knowledge graph (OWL/RDF) encoding concepts, prerequisites, and strategies.
Explainable Bayesian Updating
Interpretable probability distributions and decision rationales with each interaction.
Multi-Objective Recommendation
Balances learning gain, remediation urgency, and engagement.
Counterfactual Reasoning
Simulates alternative pedagogical sequences to justify paths.
Technology Stack
Algorithms & Methods
Item Response Theory (IRT)
Psychometric modeling for learner proficiency estimation.
Bayesian Inference
Dynamic belief updating over student mastery states.
Knowledge Graph Reasoning
OWL/RDF ontology traversal for prerequisite and dependency analysis.