Optimal Hierarchical Learning Path Design With Reinforcement Learning

E-learning systems are capable of providing more adaptive and efficient learning experiences for learners than traditional classroom settings. A key component of such systems is the learning policy. The learning policy is an algorithm that designs the learning paths or rather it selects learning materials for learners based on information such as the learners’ current progresses and skills, learning material contents. In this article, the authors address the problem of finding the optimal learning policy. To this end, a model for learners’ hierarchical skills in the E-learning system is first developed. Based on the hierarchical skill model and the classical cognitive diagnosis model, a framework to model various mastery levels related to hierarchical skills is further developed. The optimal learning path in consideration of the hierarchical structure of skills is found by applying a model-free reinforcement learning method, which does not require any assumption about learners’ learning transition processes. The effectiveness of the proposed framework is demonstrated via simulation studies.     

社会层级的内隐刻板印象：权力与地位的不同作用

采用探测词再认范式,考察社会层级的内隐刻板印象。结果发现,在能力推理上,无论地位高低,个体对高权力推理为积极的反应时显著长于消极,对低权力推理为消极的反应时显著长于积极;在热情推理上,无论权力高低,个体对高地位推理为积极的反应时显著长于消极,对低地位推理为消极的反应时显著长于积极。这表明,权力和地位在热情和能力维度的刻板印象中起着不同的作用,其中权力正向预测能力,地位正向预测热情。     

Attribute Dismissal and Valence Effects in Preferential Decision Processing

The processing of attribute information during preference‐based decision making is affected by both the valence of that information and its importance to the decision. Although these two factors have typically been examined separately, we propose that their effects on elaboration and encoding are often codependent. Results of four experiments demonstrate that the traditional negativity effect, whereby negative attribute information is processed more extensively than positive attribute information, obtains only for the subset of attributes perceived to be most important. Among other attributes, the negativity effect is reduced or even reversed (a positivity effect). Our findings suggest important qualifications to prevailing notions of selective information processing. Copyright © 2017 John Wiley & Sons, Ltd.     

认知诊断计算机化自适应测验中的项目增补

项目的增补对认知诊断计算机化自适应测验(CD-CAT)题库的开发与维护至关重要。借鉴单维项目反应理论(IRT)中联合极大似然估计方法(JMLE)的思路, 提出联合估计算法(JEA), 仅依赖被试在旧题和新题上的作答反应联合地、自动地估计新题的属性向量和新题的项目参数。研究结果表明：当项目参数相对较小且样本量相对较大时, JEA算法在新题属性向量和新题项目参数估计精度方面表现不错; 而且样本大小、项目参数大小以及项目参数初值都影响着JEA算法的表现。     

计算机化自适应诊断测验中原始题的属性标定

认知诊断测验项目开发成本较高, 要标定大量项目的属性相当费时费力, 专家完成这一任务也比较困难。对于在计算机化自适应诊断测验中的项目属性的标定尚未见到报导。在已有的为诊断测验开发的小型题库基础上, 本文在计算机化自适应认知诊断测验过程中, 植入原始题, 对项目属性标定的问题进行探讨, 重点研究原始题属性标定的方法及其影响因素, 除了MMLE方法和MLE方法外, 还建立了一种新的可用于所有非补偿认知诊断模型的属性标定的方法—— 交差方法。Monte Carlo模拟结果显示, MMLE方法较MLE方法好; 在知识状态估计精度较高时, 自适应植入原始题较随机植入原始题有一定的优势; 随着知识状态估计精度提高和原始题作答次数增加, 交差方法与MLE方法基本相当, 只是在发散型和无结构型表现欠佳, 但是交差方法不需要预先设定项目参数值。     

认知诊断测验蓝图的设计

通常认为由属性和项目关联阵（即Q矩阵）的列对应的项目充任认知诊断测验中行为样本,其实这种做法不能有效防止理想反应模式的误判。如在测验之前便可确定欲测之属性及层级关系,找到可达阵,可证明可达阵的各个列对应的项目类在认知诊断测验中必不可少,否则在理想反应模式下就一定有一些被试会被误判。本文给出充分必要Q矩阵的概念,以区别Tatsuoka(1995,2009) 讨论过的充分Q矩阵概念。充分必要Q矩阵才能有效指导测验的编制。     

STIMULUS CHARACTERISTICS WITHIN DIRECTIVES: EFFECTS ON ACCURACY OF TASK COMPLETION

Three experiments were conducted in an outpatient setting with young children who had been referred for treatment of noncompliant behavior and who had coexisting receptive language or receptive vocabulary difficulties. Experiment 1 studied differential responding of the participants to a brief hierarchical directive analysis (least‐to‐most complex stimulus prompts) to identify directives that functioned as discriminative stimuli for accurate responding. Experiment 1 identified distinct patterns of accurate responding relative to manipulation of directive stimulus characteristics. Experiment 2 demonstrated that directives identified as effective or ineffective in obtaining stimulus control of accurate responding during Experiment 1 continued to control accurate responding across play activities and academic tasks. Experiment 3 probed effects of the interaction between the type of directive (effective vs. ineffective) and the reinforcement contingency (differential reinforcement for attempts vs. differential reinforcement for accurate responses) on accurate task completion and disruptive behavior. Results suggested that behavioral escalation from inaccurate responding to disruptive behavior occurred only when ineffective directives were combined with differential reinforcement for accurate task completion. The overall results are discussed in terms of developing a methodology for identifying stimulus characteristics of directives that affect accurate responding.     

Vertical and Horizontal Transcendence

Transcendence is explored from two perspectives: the traditional concept wherein the origination of the sacred is "out there," and the alternate concept wherein the sacred originates "here." Each is evaluated from the perspectives of aesthetics and hierarchy. Both forms of transcendence are viewed as essential to the full religious life.     

小学四、五年级数学诊断性测验的编制——基于规则空间模型的方法

基于规则空间模型, 以小学四、五年级数学诊断性测验的编制为例, 探索了认知诊断理论背景下诊断性测验的编制方法。研究发现, 基于规则空间模型编制的诊断性测验具备优良的信效度, 尤其在结构效度上具有突出优势。应用该测验对1059名四、五年级学生进行诊断测验的结果显示:在整体上, 学生对整数、初级运算与应用掌握得较为巩固, 对量、统计、规律、高级运算掌握较差;在发展趋势上, 量、统计、规律、高级运算是四、五年级之间进步最快的属性。     

基于属性掌握概率的认知诊断计算机化自适应测验选题策略

在认知诊断计算机化自适应测验(CD-CAT)中, 被试对每个属性的掌握概率更直接地反映了被试能力的当前估计值。因此, 基于被试的属性掌握概率来构建选题策略, 选择最能改变被试属性掌握概率的题目作为下一个测验项目, 这应该是一个值得尝试的方案。本文借鉴已有相关研究的数据生成模式进行探索, 模拟实验结果表明：假设属性间相互独立,在定长(长度为16)、变长(长度为16或后验属性掌握模式概率达到0.8)以及短测验(长度分别为4、6、8、10)的情况下, 基于属性掌握概率的选题策略PPWKL和PHKL有较好的分类准确率, 在题目曝光率, 题库使用均匀性等方面也有较好的表现; 与研究较多的PWKL、HKL等策略相比, 也略有优势; 当属性间存在不同程度的相关时, 在定长、变长以及较短的测验条件下, 基于PHKL和MI的测验对知识状态估计精度较好, 基于PPWKL和PHKL的测验综合表现占优。