SGAASS: A Closed-Loop Adaptive Framework for Personalized Skill Gap Analysis and Deadline-Aware Study Scheduling Using NLP

Abstract

The fast-changing developments in industries present a new challenge by leaving learners' ability to meet the expectations of employers. Hence, available educational platforms keep providing learning materials that are based more on the general popularity of specific subjects than on assessing the skills lacking for fulfilling a specific job role. In this paper, we introduce the new SGAASS (Skill Gap Analyzer and Adaptive Study Scheduler), a simple-to-use framework that combines bigram-based TF-IDF skills extraction, a weighted skill-gap analysis based on cosine similarity calculation, and the new Adaptive Priority Scheduler (APS) designed to build enhanced study plans on the fly. As opposed to traditional recommending platforms, APS continually adjusts priorities using combined data on the level of skill gaps, learner's success in mastering required skills, learning speed, and urgency of deadlines and the cold start problem according to the approach of gradient-free optimization. This framework utilizes only standard CPU hardware without necessitating the use of large language models or GPU processing, thereby guaranteeing effective deployment and computational efficiency. The results from experiments on SGAASS-Bench-v1 (2,500 authentic job listings and 500 apprentice profiles) illustrate that SGAASS obtains Skill Match Accuracy of 91.4%, F1-score of 87.4%, and NDCG@10 of 89.3%, outperforming different baselines including TF-IDF, BERT, and GraphSAGE and lowering the inference time at the same time. In addition, the 4-week study involving 50 respondents demonstrated user satisfaction rating of 4.6 out of 5 and 78.4% schedule implementation ratio, hence proving efficiency of the framework in terms of personalized and deadline-conscious career preparedness and skill acquisition.