互联网、移动信息技术的快速发展使得传统的学习环境发生了巨大变革，走 向了虚拟数字化环境和现实物理环境的逐步融合。在这种混合式学习环境中，以 测量为核心的“数据驱动教学”强调在合理量化与全面收集学习者学习行为数据 的基础上，利用多种数据处理工具，测量和分析各种学习场景下产生的多维度数 据与信息，并对这些个性化学习数据进行深层挖掘、多维量化、综合分析、可视 化显示。为了优化中小学混合式学习的效果，提升学生的综合素养，以下实际问 题亟需解决：基于教育云环境，如何构建动态化的诊断与干预系统并服务于中小 学开展信息技术支持下的混合学习？ 本研究借助 Parsons 的“AGIL”模型提出“数据驱动的动态学习干预模型分 析设计框架”，同时根据问卷调查与专家访谈所得结果构建了教育云环境下数据 驱动的动态学习干预模型。基于该模型设计与建构了“动机唤醒干预”子系统、 “学习水平诊断与干预”子系统和“认知能力诊断与干预”子系统，并用于中学 教学实践活动中，对充满个性差异的学习者进行学习诊断，根据诊断结果实施针 对性的学习干预活动，提高学生的学业水平。具体内容包括： 1. 在对已有研究成果和文献的梳理与归纳的基础上，通过 Delphi 方法进行 两轮专家征询，编制了混合学习环境下学习绩效影响因素问卷，并实施调查，明 确了混合学习环境下的学习绩效影响因素。该结果可以有效指导设计学习干预模 型和环境的建立； 2. 基于问卷调查、专家访谈结果以及数据驱动的动态学习干预模型分析设 计框架，对“数据驱动的动态学习干预”开展理论构建，提出了数据驱动的动态 学习干预模型，并利用模糊综合评价的方法对该模型进了相关评价； 3. 基于数据驱动的动态学习干预模型所包含的环境因素，提出数据驱动的 动态学习干预环境的基本构成。构建了一个由电子书包平台、动机唤醒干预系统、 学习水平诊断与干预系统、认知能力诊断与干预系统有机统一的架构体系。并分 别对这几个系统的结构、功能、组成以及设计实现进行了详细的论述； 4.在真实教学场景中验证数据驱动的动态学习干预模型的有效性。开展了动 机唤醒干预实施效果验证，通过相关行为数据，利用重复测量方差分析、随机森 林算法预测分类分析技术对不同种类的行为数据进行分析，发现干预效果较好， 学生的学习行为数据较干预前有较明显的提升；利用 AMS 成就动机量表数据对 研究对象进行前后测，通过配对样本 t 检验，验证干预前后学生的动机水平具有 II 显著性的提升； 5.开展学习水平诊断与干预效果验证实验，收集了参与实验的 168 位学生进 行了干预前后的能力值数据，并进行准实验研究，借助于 SPSS 统计分析软件， 对相关数据进行了独立样本 t 检验和重复测量分析，发现通过学习水平诊断与干 预系统在进行针对性的题目推送训练后，学生的学习能力有了较为明显的提升； 6.开展认知能力诊断与干预效果验证实验，探究了 PASS 理论下四个能力指 标以及总量表分数与物理成绩之间的相关性，通过多元回归分析，调节效应分析 等方式研究了几个变量之间的影响关系。设计了准实验研究，对实验组和控制组 前后的 CAS 量表总成绩 FS 进行了独立样本 t 检验，对组内在干预前后进行了配 对样本 t 检验。同时还针对干预与否这个调节效应条件进行了调节效应验证。 通过以上一系列的研究可知，本研究在基于教育云环境下所构建的数据驱动 的动态学习干预模型和环境能够很好的应用于中学混合式学习模式中，其三级干 预体系通过三个干预子系统对学生在学习中的动机、学习水平和认知能力开展了 针对性的诊断与干预，有效地提升了学生的学习绩效，实现了学习过程基于数据 的决策与实施。 本研究主要创新点首先是将 AGIL 模型构建的思想观点以及问题解决方法 引入教育学领域并与教育学领域相关理论成果相融合，构建了数据驱动的动态学 习干预模型理论体系。其次本研究将教育云环境下动态学习干预环境与所构建的 理论相结合，形成了具有理论指导的学习干预实践体系，为研究解决如何有效地 在中学课堂中实现基于数据的、动态化的学习干预进行了理论与实践的创新。

With Internet and mobile information technology developing, traditional learning has experienced tremendous changes, moving towards the gradual integration with virtual digitization. In the mixed learning, Data-driven teaching centered on measurement emphasizes that based upon reasonable quantification and comprehensive collection of learners' behavior data, multi-dimensional data are measured in different learning scenes by using various data processing tools. Deep exploration, multidimensional quantification, comprehensive analysis and visual display are essential in dealing with the personalized data. In order to optimize the effect of mixed learning in primary and secondary schools and improve learners' comprehensive qualities, the following problem needs to be solved urgently. How to establish a dynamic diagnosis and intervention system to serve mixed learning in information technology? With the help of Parsons' AGIL model, the framework of data-driven dynamic learning intervention model was put forward. In the mean time, according to related questionnaire and expert interview, the intervention model was constructed in the educational cloud environment. The sub-systems, motivation arousal intervention, diagnosis and intervention in learning level and cognitive ability, were constructed and applied to teaching practice to diagnose learners with different personalities and implement targeted learning intervention to promote learners' academic level. The following are main studying contents: 1. On the basis of combing and summarizing the existing research results and relevant documents, I conduct two rounds of expert consultation using Delphi method, and compile a questionnaire on influencing factors of learning performance in mixed learning environment which is used to investigate and identify the influencing factors of learning performance in mixed learning environment. The results can effectively guide the design of learning intervention model and the establishment of the environment. 2. Based on the survey results of questionnaires, expert interviews and AGIL theoretical system proposed by Parsons, the analysis and design framework of data IV driven dynamic learning intervention model is established. Based on the theoretical construction of data-driven dynamic learning intervention, a data-driven dynamic learning intervention model is proposed which is verified by fuzzy comprehensive evaluation. 3. Based on the environmental factors included in the data-driven dynamic learning intervention model, the researcher proposes the basic composition of the environment of the data-driven dynamic learning intervention. A unified framework system is constructed by e-book package platform, motivation arousal intervention system, learning level diagnosis and intervention system, and cognitive ability diagnosis and intervention system. After that, the structure, function, composition and design of these systems are discussed in detail. 4. The effectiveness of the data-driven dynamic learning intervention model in real teaching was verified. The implementation effect of motivation arousal intervention detected obvious improvement in learning behavior compared with the data before intervention by repeated measurement variance and random forest for classification. In addition, AMS data was adopted to test the subjects before and after the intervention, and paired sample T test verified the fact that the students' motivation level develops dramatically after the intervention. 5. It carried out the confirmatory experiment of diagnosis and intervention effect in learning level. The competence data of 168 subjects before and after the intervention were collected, and a quasi-experimental study was also included. Independent sample T test and paired sample T test were adopted to find that the dramatic improvement in learners’ competence after targeted training by diagnosis and intervention system in learning level. 6. The experiment of cognitive ability diagnosis and intervention effect verification is carried out to explore the correlation between four ability indexes, scale scores and physical achievements under PASS theory, and study the influence relationship between the variables by means of multiple regression analysis and adjustment effect analysis. A quasi-experiment is designed. The independent sample T test is conducted to test the FS of CAS scale in the experimental group and the control group and the matched sample T test is conducted in the group before and after the V intervention. At the same time, it also verifies the adjustment effect on the condition of intervention or not, and explores the effect of intervention on the relationship between cognitive ability FS and physical performance. From the above, the intervention model of data-driven dynamic learning based on education cloud can be applied to mixed learning in middle school effectively, and three sub-levels of which implemented targeting diagnosis and intervention in motivation, learning level, and cognitive ability to boost learners’ performance and realize the decision and program in the learning processes based on data. The following are innovations of the study. First, the perception and solution of AGIL model were applied to pedagogy, integrating with the theory of RTI and Marzano’s Taxonomy of Cognitive Objectives, and constructing theoretical system of the intervention model of data-driven dynamic learning. Second, combined with established theory in the intervention model, it developed practical intervention system with theoretical guidance concerning how to realize dynamic learning intervention on data.