A Theoretical AI - Fuzzy Logic Framework for Student Learning Profiles and Differentiated Instruction in Primary Education

Abstract

This theoretical paper presents an AI-supported fuzzy logic framework for classroom grouping and individualized instruction in primary education. The learner profile integrates three dimensions elicited via a chatbot using a three-point scale: VARK, learning preferences, and interests. Student responses are mapped to fuzzy membership values to construct a compact learner vector enabling dynamic, non-rigid classification. A Mamdani inference mechanism aggregates IF–THEN rules to generate graded recommendations across five pedagogical approaches: Project-Based Learning, Challenge-Based Learning, STEAM, Makerspace/Experiential–Collaborative Learning, and Game-Based Learning. Based on these recommendations, a grouping module supports the formation of heterogeneous or homogeneous student teams aligned with instructional goals. A simulation across varying class sizes indicates that the framework maintains low grouping time while preserving higher group-quality indicators compared to random and manual grouping strategies. The main contributions include a lightweight learner-profile model, fuzzy pedagogical recommendations, and a flexible grouping workflow complementing teacher judgment.