基于深度学习的知识追踪研究综述

A review of the progress of knowledge tracking research based on deep learning

摘要:

随着人工智能与教育的不断发展，知识追踪在智慧教学领域具有广阔的应用前景。而深度学习以其强大特征提取能力广泛应用于知识追踪，以深度学习知识追踪模型为起点，其改进模型为主线，全面回顾了知识追踪模型的研究进展，简要介绍了知识追踪领域传统模型的特点及不足，阐述了基于深度学习知识追踪模型的原理及局限性，同时全面整理并分析了针对可解释性问题、缺少学习特征、记忆增强网络、图神经网络、基于注意力机制五个方面的改进模型，梳理了知识追踪领域常用的公开数据集、评价指标及模型性能对比分析，最后总结并探讨了知识追踪在智慧教学方面的应用以及当前该研究领域的研究现状与未来的研究方向。

Abstract:

With the continuous development of artificial intelligence and education, knowledge tracking has broad application prospects in the field of smart teaching. Deep learning is widely used in the field of knowledge tracking with its powerful feature extraction ability, taking the deep learning knowledge tracking model as the starting point, and its improved model as the main line, comprehensively reviewing the research progress of the knowledge tracking model, briefly introducing the characteristics and shortcomings of the traditional model in the field of knowledge tracking, expounding the principles and limitations of the knowledge tracking model based on deep learning, and comprehensively sorting out and analyzing the problems of interpretability, lack of learning features, memory enhancement networks, graph neural networks, Based on the improvement model of five aspects of attention mechanism, the comparative analysis of public data sets, evaluation indicators and model performance commonly used in the field of knowledge tracking is sorted out, and finally the application of knowledge tracking in smart teaching and the current research status and future research direction of this research field are summarized and discussed.