A New Methodology for Clustering of Online Learning Resources Based on Students' Learning Styles

Abstract

Recommending learning resources to students according to their respective learning style is conductive to improving learning efficiency, and the clustering and classification of learning resources conduce to reducing learning resource redundancy and duplication and increasing learning resource utilization. However, available methods of learning resource recommendation usually regard students’ learning styles as fixed and invariable, which apparently contradicts reality and may have a negative influence on students’ learning effect. In view of this matter, this paper aims to propose a novel methodology for clustering online learning resources based on student learning style. At first, the specific steps of the new clustering method were given, and a Sharpe model was adopted to analyze the invalid exposure of students’ learning effect and identify students’ learning styles. The learning style coefficient of students was regarded as a dynamic systemic state, which was estimated by the Kalman filter. Then, the Affinity Propagation Clustering (APC) algorithm was adopted to cluster learning resources based on a student learning style distance matrix, and a model of recommended online learning resource combinations was established based on the proposed method. At last, experimental procedures, including learning style evaluation, pretest exam score prediction, posttest exam score prediction, and data analysis, were described in detail, and the validity of the proposed method was verified by experimental results.