CIF21 DIBBs: Building a Scalable Infrastructure for Data-Driven Discovery and Innovation in Education

CIF21 DIBB：为数据驱动的教育发现和创新构建可扩展的基础设施

• 批准号: 1443068
• 负责人: Ken Koedinger
• 金额: $ 483.08万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-01-01 至 2020-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1443068&HistoricalAwards=false
• 关键词: CIF21; DIBBs; Building; Scalable; Infrastructure

This project is creating a community software infrastructure, called LearnSphere, that supports sharing, analysis, and collaboration across the wide variety of educational data. LearnSphere supports researchers as they improve their understanding of human learning. It also helps course developers and instructors improve teaching and learning through data-driven course redesign. The goal is to transform learning science and engineering through a large, distributed data infrastructure, and develop the capacity for course developers, instructors, and learning engineers to make use of it.LearnSphere maintains a central store of metadata about what datasets exist, but also has distributed features allowing contributors control over access to their own data. It provides a hub to link many communities of educational researchers, provides a repository for researchers to store their data, and provides an open analytic method library and workflow-authoring environment for researchers to build models and run them across datasets.The research team has extensive experience not only in using educational data mining to make discoveries and improve student outcomes, but also in the creation of educational data infrastructures. They have developed the DataShop infrastructure, which is currently the largest open repository of educational technology data including over 550 datasets. A newer data infrastructure, MOOCdb, is being developed to store and analyze Massively Open Online Course (MOOC) data. The Open Learning Initiative has produced data stored in DataShop for many years and is expanding into the MOOC space. Dialogue-based tutoring systems and student affect sensors are producing new kinds of data that are being added to LearnSphere. The researchers are further improving data collection infrastructure in MOOCs especially by adding platform components for massive multi-factor online experiments. The project is also creating new methods for data integration, discourse data storage and analytics, and new algorithms for automated discovery, as well as new learning science discoveries that result from these algorithms. By integrating these building blocks in LearnSphere, the project will facilitate cross-modality and cross-domain educational data analysis that is not possible today.

该项目正在创建一个名为LearnSphere的社区软件基础设施，支持各种教育数据的共享，分析和协作。 LearnSphere支持研究人员提高他们对人类学习的理解。 它还帮助课程开发人员和教师通过数据驱动的课程重新设计来改善教学和学习。 目标是通过大型分布式数据基础设施来转变学习科学和工程，并开发课程开发人员、教师和学习工程师利用它的能力。LearnSphere维护有关存在哪些数据集的元数据的中央存储，但也具有分布式功能，允许贡献者控制对自己数据的访问。它提供了一个连接许多教育研究人员社区的枢纽，为研究人员提供了一个存储库来存储他们的数据，并为研究人员提供了一个开放的分析方法库和工作流创作环境，以构建模型并在数据集上运行它们。研究团队不仅在使用教育数据挖掘进行发现和提高学生成绩方面拥有丰富的经验，还包括创建教育数据基础设施。他们开发了DataShop基础设施，这是目前最大的开放式教育技术数据库，包括550多个数据集。一个新的数据基础设施MOOCdb正在开发中，用于存储和分析大规模开放在线课程（MOOC）数据。开放式学习计划已经在DataShop中存储了多年的数据，并正在扩展到MOOC领域。基于对话的辅导系统和学生情感传感器正在产生新类型的数据，这些数据正在被添加到LearnSphere中。研究人员正在进一步改善MOOC的数据收集基础设施，特别是通过为大规模多因素在线实验添加平台组件。该项目还创建了数据集成，话语数据存储和分析的新方法，自动发现的新算法，以及从这些算法中产生的新的学习科学发现。通过将这些构建模块集成到LearnSphere中，该项目将促进跨模式和跨领域的教育数据分析，这在今天是不可能的。

项目成果

Deep Neural Model Inspection and Comparison via Functional Neuron Pathways

• DOI: 10.18653/v1/p19-1575
• 发表时间: 2019-07
• 作者: James Fiacco;Samridhi Choudhary;C. Rosé