Tailoring the Digital Campus: Design and Semester-Long Field Evaluation of an Adaptive User Interface in Moodle LMS

Abstract

Most university learning platforms continue to serve identical, rigid menus and resource listings to all students regardless of how quickly they learn or what study habits they show. This uniform setup often creates needless friction, causing students to lose time searching for materials instead of focusing on their coursework. In this study, we built and tested an Intelligent User Interface (IUI) that plugs directly into an active university Moodle environment without changing core system files. The platform pairs a recommendation pipeline—blending collaborative filtering via Singular Value Decomposition with text-based content similarity—with an adaptive dashboard and responsive drawer navigation. Over a 16-week term, we tracked 520 university students and gathered both continuous background clickstream logs and mid-term survey feedback on usability (SUS), satisfaction (ACSI), and engagement (UWES-S). Using partial least squares structural equation modeling (PLS-SEM) and 5,000 bootstrap resamples, we evaluated how the interface influenced study habits and final grades. Recommendation tests showed strong ranking accuracy (Precision@5 = 0.842, NDCG@5 = 0.865) while adding only 41.8 ms in server processing time. Structural models indicated that interface interaction directly improved student engagement (β = 0.438, p < 0.001) and final marks (β = 0.285, p < 0.001), explaining 44.8% of variance in engagement and 35.6% of final course marks. Mediation analysis revealed that student engagement served as a substantial partial bridge (indirect effect β = 0.142, accounting for 33.25% of total effect). These findings show that universities can meaningfully upgrade existing Moodle setups with lightweight adaptive plugins that ease navigation, keep students involved, and improve academic performance.