网络学习空间中学习画像的标签模型构建研究

Research on Building Label Model of Learning Profile in Network Learning Space

混合式学习是互联网时代的一种主要学习形式。学生在网络学习空间中的学习活动将会产生大量的学习过程和学习结果数据。挖掘这些数据的价值成为了教育信息化的热点问题。基于这些学习数据开展学生学习画像将有助于分析学生的学习行为和学习成效,为探索个性化教育提供充分的数据支撑。该文以学习画像中最典型的学习能力和学习行为偏好两个维度的分析刻画为例,根据学习数据的动态周期性特征,提出了一种调整的线性加权变异系数算法,消除了量纲与样本容量的影响,实现了学生学习能力稳定性标签模型;同时利用箱线图k百分位数方法,结合偏好随机变量概率分布理论,构建了学生行为偏好中最有代表性的学习任务响应习惯标签模型。通过对采集的样本数据和处理结果的对比分析表明,采用该方法构建的两种标签模型具有良好的刻画效果,准确地反映了学生的个体学习特征。

Blended learning is a major learning method in the age of network.Students’learning activities in network learning space will produce a large amount of learning process and learning result data.Mining the value of these data has become a hot issue of education informatization.Portraying student learning based on these learning data will be helpful to analyze students’learning behavior and learning effectiveness,and then provide sufficient data support for exploring personalized education.Learning ability and learning behavior preference are the most typical dimensions in the learning profile.Taking the two dimensions of learning ability and learning behavior preference as an example,according to the dynamic periodic characteristics of learning data,we propose an adjusted linear weighted coefficient of variation algorithm,which eliminates the influence of dimension and sample size,and realizes the stability label model of students’learning ability.Using boxplot,k-percentile and probability distribution theory of preference random variables,the most representative learning task response habit label model is constructed.By comparative analyzing of the collected sample data and processing results,the two label models have excellent characterization effects,and accurately reflect the individual learning characteristics.