Goal Aware Fine Grained Personalized Learning Framework Adaptive to Dynamic User Profile

Abstract

Personalized e-learning platforms enhance the learning experience by adapting educational content according to learners' learning styles, knowledge levels, and affective states. In our previous work, a fine-grained learning style prediction framework was proposed that adapted to learner knowledge profiles, content semantics, and temporal affective states. However, the framework relied on coarse-grained interaction statistics to model learner preferences, limiting its ability to capture the temporal evolution of learner behavior during the learning process. Furthermore, the framework personalized learning resources solely based on learner characteristics without considering administrator-defined learning objectives. To address these limitations, this paper proposes a Goal-Aware Fine-Grained Personalized Learning Framework (GAFG-PLF) that jointly models content semantics, learner knowledge profiles, temporal affective states, dynamic interaction behavior, and administrator-defined learning goals. The proposed framework consists of two key components: (i) a Dynamic Preference Evolution Network, which continuously estimates evolving learner preferences using multimodal learning cues, and (ii) a Goal-Aware Recommendation Engine, which recommends the most appropriate learning resources by integrating predicted learning styles, dynamic learner preferences, knowledge profiles, affective states, and instructional objectives. Experimental evaluation demonstrates that the proposed framework significantly improves learning style prediction accuracy and enhances recommendation quality compared with existing approaches. These improvements contribute to a more adaptive, learner-centric, and goal-oriented personalized e-learning environment, thereby supporting better learning outcomes and more effective educational resource delivery.