Designing an Intelligent Adaptive Learning Assistant Using Natural Language Processing to Enhance Students' Academic Performance

Abstract

The research paper substantiates and empirically tests an intelligent adaptive learning assistant that combines Natural Language Processing (NLP), mastery-based adaptive sequencing, and real-time learning analytics to improve academic output in higher education. The architecture of the system is a modular architecture, which is comprised of an NLP processing pipeline of intent recognition and concept mapping, an adaptive engine of dynamic feedback generation based on the estimation of mastery, and an analytics dashboard for monitoring the instructor and decision support. A quasi-experimental study was carried out on 120 undergraduate students in experimental (n = 60) and control (n = 60) groups and lasted four weeks of intervention. The evaluation indicators were pre- and post-test outcomes, interaction records, and retention levels. Results demonstrated statistically significant improvements in learning outcomes for the experimental Group, with a post-test mean score of 81.5 compared to 69.3 in the control group (p < 0.001). The normalized learning gain increased from 6.9 to 18.4 (+166.6%), retention rates improved from 83.5% to 94.1%, and time-on-task was reduced by 14.7%, indicating enhanced learning efficiency. The significant instructional influence of the proposed system was proven by a large effect size (Cohen d = 1.19). These results confirm that the adaptive feedback processes that are fueled by NLP can provide a meaningful bridging between static digital learning conditions and customized teaching interventions, with the resultant cognitive and behavioral improvements, as well as the opportunity to make pedagogical decisions that are informed by analytics.