PROPOSAL FOR AI-DRIVEN ADAPTIVE E-LEARNING SYSTEM FOR PERSONALIZED STEM KNOWLEDGE ASSESSMENT
Keywords:
Аdaptive e‑learning, artificial intelligence, Bayesian Knowledge Tracing, deep knowledge tracing, intelligent tutoring systems, personalized STEM educationAbstract
Modern STEM-focused e-learning systems often overlook nuanced, individual-level knowledge gaps in learners, hindering their preparedness for advanced topics. We propose a proof-of-concept, AI-driven adaptive e-learning platform that combines Bayesian Knowledge Tracing (BKT) and advanced recurrent neural network (RNN)-based models—such as Deep Knowledge Tracing—to accurately model learners’ evolving conceptual mastery . The system continuously processes interaction data—including correct/incorrect responses, timestamps, hint requests, and error types—to calculate mastery probabilities for each STEM concept in real-time. BKT provides transparent, interpretable latent-state updates governed by key parameters (p_init, p_transit, p_slip, p_guess), while RNN models (e.g., LSTM/GRU variants) enhance predictive accuracy by capturing complex temporal dependencies and long-term learning patterns.
Upon identification of knowledge gaps, the platform employs a recommendation engine—potentially powered by reinforcement learning or rule-based heuristics—to dynamically generate personalized learning paths and targeted exercises. This approach follows successful methodologies like the Attentive Knowledge Tracing system, which has demonstrated measurable improvements in learning outcomes and path diversity. The architecture utilizes a microservice framework, with a Python backend (Django/Flask), a JavaScript front end, and RESTful AI services facilitating real-time adaptation.
Anticipated outcomes include quantitative metrics on model performance (e.g., precision/recall for concept mastery detection), as well as simulated evaluation of recommendation efficacy using synthetic or historical datasets. This research builds on evidence-based strengths of BKT and deep learning in tracking learner knowledge, while also advancing adaptive e-learning through its hybrid design.
By enabling early and accurate identification of learning gaps and providing individualized paths to address them, the proposed platform represents a scalable, modular advancement in Intelligent Tutoring Systems. It targets improved learner progression through complex STEM competencies and demonstrates the transformative potential of AI-supported educational tools.
References
Baker, R. S. J. d., & Inventado, P. S. (2014). Educational data mining and learning analytics. In Learning analytics (pp. 61–75). Springer. https://doi.org/10.1007/978-1-4614-3305-7_4
Corbett, A. T., & Anderson, J. R. (1995). Knowledge tracing: Modeling the acquisition of procedural knowledge. User Modeling and User-Adapted Interaction, 4(4), 253–278. https://doi.org/10.1007/BF01099821
Chen, J.-Y., Saeedvand, S., & Lai, I.-W. (2023). Adaptive learning path navigation based on knowledge tracing and reinforcement learning. arXiv. https://arxiv.org/abs/2305.04475
Liu, V., Latif, E., & Zhai, X. (2025). Advancing education through tutoring systems: A systematic literature review. arXiv. https://arxiv.org/abs/2503.09748
Piech, C., Bassen, J., Huang, J., Ganguli, S., Sahami, M., Guibas, L. J., & Sohl-Dickstein, J. (2015). Deep knowledge tracing. In Advances in Neural Information Processing Systems (pp. 505–513). https://arxiv.org/abs/1506.05908
Piech, C., Sahami, M., & Guibas, L. (2016). Learning recommendation strategies for adaptive tutoring systems. Proceedings of the 25th International Conference on World Wide Web, 1133–1143. https://doi.org/10.1145/2872427.2883005
Wang, Y., Heffernan, N., & Heffernan, C. (2017). Deep knowledge tracing with attention-based recurrent neural networks. In Proceedings of the 10th International Conference on Educational Data Mining (pp. 74–83). International Educational Data Mining Society.
Zanellati, A., Di Mitri, D., Gabbrielli, M., & Levrini, O. (2023). Hybrid models for knowledge tracing: A systematic literature review. IEEE Transactions on Learning Technologies, 16(5), 569–584. https://doi.org/10.1109/TLT.2023.3348690
Zhu, M., & Liu, X. (2021). Reinforcement learning based recommendation systems for personalized education. Journal of Educational Technology & Society, 24(2), 123–134. https://doi.org/10.2307/2691736
Zhou, Y., Li, X., Cao, Y., Zhao, X., Ye, Q., & Lv, J. (2021). LANA: Towards personalized deep knowledge tracing through distinguishable interactive sequences. arXiv. https://arxiv.org/abs/2105.06266
