基於知識追蹤與強化式學習之適性化學習路徑推薦系統

Abstract

本研究提出了一個基於知識追蹤與強化式學習的適性化學習路徑推薦系統，旨在提供更有效的學習體驗。透過結合知識追蹤模型與強化式學習演算法，我們的系統能夠對學生的學習狀態進行精確評估，從而為每位學生設計最佳的學習路徑，以符合其個別的學習需求和能力水平。本系統的實驗結果顯示，我們的適性化學習路徑推薦能有效地幫助學生高效地達成學習目標。針對知識追蹤任務中常見的資料不平衡問題，本研究提出了一種創新的資料去重複方法，有效提高了模型的學習診斷效能。學習路徑的生成，則是採用深度強化式學習演算法來實現。為了進一步提升系統的適應性和可靠性，本論文引入了虛擬學生的概念。通過模擬大量虛擬學生數據，本系統能夠對學習路徑推薦策略進行有效的優化，從而提升系統的整體性能和穩定性。此方法不僅提高了教學模型的適應性，也為未來教育科技應用提供了新的研究方向和可能性。This research presents an adaptive learning path recommendation system based on knowledge tracing and reinforcement learning to enhance learning experiences. By integrating knowledge tracing models with reinforcement learning algorithms, our system accurately assesses student learning states and designs optimal learning paths tailored to individual needs. Experimental results show that our recommendations effectively help students achieve their learning goals efficiently.To address data imbalance in knowledge tracing tasks, we introduce an innovative data deduplication method that improves model performance. Learning paths are generated using deep reinforcement learning algorithms.We also introduce the concept of virtual students to simulate data, optimizing learning path recommendations and improving system performance and stability. This approach enhances the adaptability of the instructional model and opens new research directions for educational technology applications.