以慕课的持续优化为应用目标，充分考虑慕课学习者在基础和动机等方面的差异性，将课程学习材料、课程学习活动数据、课程学习者都看作是相关教育资源的不同部分，研究（1）教育资源的语义化（标注）及其匹配；（2）针对学习进展的教育资源之间的语义关联；（3）从先修基础到学习目标之间学习路径的存在性、可行性，以及学习材料之间的一致性与相关性。探索在线教育资源组织的理论，提出基于课程的在线教育资源优化标准框架及其动态评估方法，提高对慕课效用的综合认识，为慕课在执行过程中以及重开之间的改进提供指导。研究工作将基于实际运行的慕课平台展开，采用用户问卷、数据采集、A/B实验和模型分析等方法。在过去两年里，参与本项目申请的三个单位都开设了若干慕课，获得了大量学习行为数据，同时也管理运行着大型慕课平台。这样的基础条件不仅使我们能不断获得研究所需的学习行为大数据，而且也利于研究工作成果的实现与应用。

We propose a set of innovative theory and methods to integrate education resources in Massive Open Online Courses (MOOCs) in order to improve the quality of MOOCs over time. We argue that the MOOC resources should not only include course contents and learning data, but also the human resource, i.e., the teachers and the learners themselves. We focus on the three key research problems: 1) automatic semantic labeling of all MOOC resources; 2) discovery of the semantic links among resources, especially those links that affect the learning process; and 3) analysis of the existence and reachability of the learning path for each individual student, from prerequisites all the way to the learning goal. Based on our theoretical work on educational resource organization, we propose dynamic evaluation methods and metrics, which can be used to improve MOOC operation over time. Each of three collaborating institutions operates its own large scale MOOC platform, and has developed a number of successful MOOCs over the past two years. Our research is based on experience and learning data from these MOOC practice. These real platforms will serve as a living laboratory for proposed research projects, allowing us to perform user studies with surveys and A/B testing in addition to passive data collection. Last but not least, we will apply our research directly on these real platforms and benefit millions of learners.