AI-Enhanced Curriculum Design and Deep-Learning-Based Assessment in International Sports Communication Education

Abstract

Traditional assessment in international sports communication is often fragmented and subjective, limiting timely, learner-centered feedback. This study presents a curriculum framework enhanced by generative artificial intelligence, coupled with a deep learning (DL) model for instructional effectiveness assessment in international sports communication. The pipeline integrates de-identified learning analytics—learning management system clickstreams, interaction networks, and rubric-scored artifacts—into engineered features for DL training with parameter search and cross-validation. A 16-week field study across three undergraduate sections at a Chinese comprehensive university (N = 60; two involving generative artificial intelligence, one comparison) benchmarked DL against linear regression and decision tree baselines and against expert ratings on intercultural communication competence, framing diversity, and production quality. Results show that DL converged faster and yielded lower prediction error than the baselines, while closely aligning with expert scores, enabling actionable, personalized feedback and course tuning driven by constructive alignment.