[论文解读] Bi-CLKT: Bi-Graph Contrastive Learning based Knowledge Tracing

Xiangyu Song, Jianxin Li+4 more|arXiv (Cornell University)|Jan 22, 2022

Online Learning and Analytics被引用 4

一句话总结

该论文提出Bi-CLKT，一种新颖的知识追踪模型，通过双图对比学习联合学习练习和概念表征，利用'练习到练习'（E2E）和'概念到概念'（C2C）关系子图。通过应用节点级和图级对比学习，并结合联合损失函数整合表征，Bi-CLKT在四个真实世界数据集上实现了最先进性能，准确率和AUC最高提升3%。

ABSTRACT

The goal of Knowledge Tracing (KT) is to estimate how well students have mastered a concept based on their historical learning of related exercises. The benefit of knowledge tracing is that students' learning plans can be better organised and adjusted, and interventions can be made when necessary. With the recent rise of deep learning, Deep Knowledge Tracing (DKT) has utilised Recurrent Neural Networks (RNNs) to accomplish this task with some success. Other works have attempted to introduce Graph Neural Networks (GNNs) and redefine the task accordingly to achieve significant improvements. However, these efforts suffer from at least one of the following drawbacks: 1) they pay too much attention to details of the nodes rather than to high-level semantic information; 2) they struggle to effectively establish spatial associations and complex structures of the nodes; and 3) they represent either concepts or exercises only, without integrating them. Inspired by recent advances in self-supervised learning, we propose a Bi-Graph Contrastive Learning based Knowledge Tracing (Bi-CLKT) to address these limitations. Specifically, we design a two-layer contrastive learning scheme based on an "exercise-to-exercise" (E2E) relational subgraph. It involves node-level contrastive learning of subgraphs to obtain discriminative representations of exercises, and graph-level contrastive learning to obtain discriminative representations of concepts. Moreover, we designed a joint contrastive loss to obtain better representations and hence better prediction performance. Also, we explored two different variants, using RNN and memory-augmented neural networks as the prediction layer for comparison to obtain better representations of exercises and concepts respectively. Extensive experiments on four real-world datasets show that the proposed Bi-CLKT and its variants outperform other baseline models.

研究动机与目标

• 解决现有知识追踪模型忽视练习层面差异且无法建模复杂节点关系的局限性。
• 通过自监督对比学习整合练习和概念表征，提升预测性能。
• 通过利用高阶关系结构，克服跨概念练习中的数据稀疏性和模型不准确性。
• 在统一的对比学习框架中探索不同增强、嵌入和预测机制的有效性。
• 开发统一模型，以捕捉局部（练习级）和全局（概念级）语义信息，实现更优的知识追踪。

提出的方法

• 模型构建两个子图：基于学生交互模式的'练习到练习'（E2E）子图和'概念到概念'（C2C）子图。
• 在E2E子图上应用节点级对比学习，以学习具有区分性的练习表征。
• 在C2C子图上应用图级对比学习，以学习具有区分性的概念表征。
• 设计联合对比损失，联合优化E2E和C2C表征，增强特征区分能力。
• 实现两种变体：Bi-CLKT-R在预测层使用循环神经网络，Bi-CLKT-M在预测层使用记忆增强网络。
• 通过拼接E2E和C2C嵌入形成练习表征，进而用于最终学生表现的预测。

实验结果

研究问题

• RQ1 双图对比学习是否能通过联合建模练习和概念关系来提升知识追踪？
• RQ2 不同增强策略（如度中心性、PageRank中心性）如何影响模型在不同数据集上的泛化能力？
• RQ3 结合E2E和C2C嵌入是否比单独使用任一嵌入获得更优性能？
• RQ4 在知识追踪中，预测机制选择——RNN还是记忆增强网络——是否导致更优性能？
• RQ5 联合对比损失在多大程度上提升了表征学习和下游预测准确率？

主要发现

• Bi-CLKT-R在ASSISTment 2009数据集上达到0.875的AUC和0.802的准确率，优于所有基线模型至少3%。
• 采用PageRank中心性增强的模型在四个数据集中的三个上表现最佳，包括在STATICS 2011数据集上达到0.865的AUC。
• 拼接E2E和C2C嵌入显著优于单独使用任一嵌入，其在ASSISTment 2009数据集上达到最高AUC 0.875。
• Bi-CLKT-R在三个数据集上的表现优于Bi-CLKT-M至少2%，尤其在ASSISTment 2009和STATICS 2011数据集上表现突出。
• 消融实验证实，联合对比损失和多层级图学习显著提升了表征质量和预测准确率。
• 模型展现出强大的泛化能力，不同增强方法（度中心性 vs. PageRank中心性）在多样化数据集上均表现良好，且未出现过拟合。