Research on Adaptive Learning Path Optimization of Online Courses for Business Administration Major in Higher Vocational Education Based on Deep Reinforcement Learning

Abstract

Online courses in higher vocational business administration require learning-path optimization that can adapt to heterogeneous knowledge foundations, learning behaviors, and vocational competency objectives. To overcome the static and one-size-fits-all limitations of conventional online teaching paths, this study proposes an adaptive learningpath optimization model based on deep reinforcement learning. Course knowledge points, prerequisite relations, jobcompetency requirements, student learning behaviors, and personalized profiles are collected and transformed into a structured learning data sample library. A DQN-based decision model is then constructed, where the state space includes knowledge mastery, learning progress, learning fatigue, and career fit, and the action space includes knowledge-point sequencing, resource-type selection, and learning-time allocation. A multi-dimensional reward function integrates knowledge improvement, learning efficiency, and vocational adaptability. Experiments using the online courses “Fundamentals of Management” and “Marketing” over one semester show that the experimental group improves course completion rate by 18.3%, knowledge-point mastery by 21.5%, and learning satisfaction by 15.7% compared with the fixed-path control group. The model provides a data-driven decision-optimization method for adaptive learning systems and intelligent educational resource scheduling.