面向学习者个性化学习特征的知识追踪  

Knowledge tracing for personalized learning features of learners

作　　者：桂妍 秦学 GUI Yan;QIN Xue(College of Big Data and Information Engineering,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学大数据与信息工程学院,贵州贵阳550025

出　　处：《现代电子技术》2025年第9期104-108,共5页Modern Electronics Technique

基　　金：贵州省第二批省级金课《大数据原理与技术》线上线下混合式课程(2024JKHH0033)建设成果。

摘　　要：针对现有知识追踪模型未能充分建模学习者个体学习特征的问题,提出一种面向学习者个性化学习特征的知识追踪模型。首先,基于学习者的交互记录构建一个多维度的问题表征体系;接着,通过分析学习者的个体特性及其与问题的知识差距,实现对学习收益的个性化建模;此外,考虑到学习过程中知识的遗忘现象,设计了一个遗忘门动态调整知识的遗忘程度;最后,通过融合学习收益和知识遗忘的情况更新学习者的知识状态,并利用此状态预测学习者的未来答题表现。在两个公开在线教育数据集ASSISTments2012和ASSISTments2017上的实验结果显示,相较已有主流模型,该模型取得了更好的预测性能,能够更好地建模学习者的知识状态。Since the existing knowledge tracing models are incapable to adequately model personalized learning characteristics of learners,a knowledge tracing model oriented towards personalized learning characteristics is proposed.Initially,a multi-dimensional question representation system is constructed based on learners′interaction records.Following this,a personalized modeling of learning benefits is achieved by analyzing learners′personalized characteristics and their knowledge gaps.Additionally,a forgetting gate is designed to dynamically adjust the degree of forgetting of knowledge because of the knowledge forgetting during the process of learning.Finally,the learners′knowledge states are updated by integrating learning benefits and knowledge forgetting,and this state is used to predict learners′future performance in answering questions.Experimental results on two public online educational datasets,named ASSISTments2012 and ASSISTments2017,show that the proposed model achieves better predictive performance and can model learners′knowledge states more effectively in comparison with the existing mainstream models.

关 键 词：个性化学习 知识追踪 学习收益 学习遗忘 学习特征 教育数据挖掘 

分 类 号：TN919-34[电子电信—通信与信息系统] TP391.1[电子电信—信息与通信工程]