摘要
学习状态是决定学习质量和学习效率的重要因素。个性化学习系统可以从学习状态出发,实时监控学生的学习状态并推荐个性化的学习策略。目前,学习状态的研究大多基于问卷调查或教师的主观判断,这些方法忽视了学生学习时产生的客观数据。基于此,文章提出一种基于托普利兹逆协方差的学习状态挖掘算法,将高维时间序列分割为时序时间线,使用多层马尔可夫随机场定义每个聚类,以时间一致的方式通过动态规划算法将全局问题分解为多个子问题进行求解。实验结果表明,该算法具有较高的计算速度和稳定性,适用于线上教学平台的多维时间序列应用场景,有助于教师了解学生的学习状态,从而适时地调整教学策略和指导方法。
The learning status is an important factor that determined the quality and efficiency of learning.The personalized learning system could monitor learners’learning status in real time and recommend personalized learning strategies starting from learning status.The current studies of learning status are mostly based on questionnaire surveys or teachers’subjective judgments,while these methods ignored the objective data generated by learners when they study.This paper proposed a mining algorithm of learning status based on Toeplitz inverse covariance,which divided high-dimensional time series into time series timelines,used multi-layer Markov random fields to define each cluster,and decomposed the global problem into multiple sub-problems for solution through dynamic programming algorithm in a consistent manner.The experiments demonstrated that the algorithm had high calculation speed and stability,and was suitable for multi-dimensional time series application scenarios of online teaching platforms,which helped teachers understand students’learning status and adjust teaching strategies and guidance methods.
作者
张逸谦
王志强
梁正平
ZHANG Yi-qian;WANG Zhi-qiang;LIANG Zheng-ping(College of Computer Science and Software Engineering,Shenzhen University,Shenzhen,Guangdong,China 518060)
出处
《现代教育技术》
CSSCI
2021年第8期119-126,共8页
Modern Educational Technology
基金
2018年度深圳市科技计划资助项目“深圳市多媒体与虚拟现实公共技术服务平台”(项目编号:GGFW2018020518310863)的阶段性研究成果。
关键词
多维时间序列聚类
层次分析法
学习状态
个性化学习
multidimensional time series clustering
analytic hierarchy process
learning state
personalized learning
