Comparing Accuracy of Bidirectional Encoders Through Analysis of Computer Science Student Skillsets for Early Detection of Student Competencies and Weaknesses
Skill assessment of students is a critical process that evaluates their proficiency, ranging from novice to mastery, in various domains. This paper aims to develop a framework for predicting students’ multi-skillsets effectively.
The global IT industry recognizes that for computer science students to excel, they must acquire and demonstrate proficiency in Soft Skills, Life Skills, and Technical Skills. This study addresses the need for skill-based education in software engineering to bridge the gap between academic learning and industry requirements.
The study employs the Bidirectional Encoder Representations from Transformers (BERT) model to predict students’ competencies in Soft, Life, and Technical skills. The effectiveness of the model is evaluated using key performance indicators, including precision, accuracy, F1-score, recall, and the area under the curve (AUC), as discussed in the paper.
The paper introduces a multi-skillset framework that leverages AI techniques to predict students’ skill levels, ensuring better alignment between academic training and industry expectations.
The proposed BERT model achieves a superior prediction accuracy of 96.25%, outperforming conventional machine learning baselines.
Educators and institutions should integrate AI-driven skill assessment models to personalize learning pathways and enhance students’ industry readiness.
Further exploration of deep learning techniques can refine skill prediction models, leading to more robust assessment frameworks.
By improving the accuracy of skill assessments, this study contributes to producing future-ready software engineering professionals, ultimately enhancing workforce competency in the IT industry.
Future research can focus on integrating BERT with hybrid deep learning models, such as BERT-GRU or BERT-CNN, to enhance the accuracy and robustness of predictions. Additionally, the model can be expanded to incorporate additional skill domains and be adapted for different educational disciplines.
Back