Collaborative Research: Frameworks: Cyber Infrastructure for Shared Algorithmic and Experimental Research in Online Learning

协作研究：框架：在线学习中共享算法和实验研究的网络基础设施

• 批准号: 1931419
• 负责人: Ryan Baker
• 金额: $ 140万
• 依托单位: University of Pennsylvania
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-10-01 至 2024-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1931419&HistoricalAwards=false
• 关键词: Collaborative; Research; Frameworks; Cyber; Infrastructure

This project, RAILKaM, will create new technology that will enable twenty researchers during the grant period to run large-scale field experiments where they study basic principles in education and educational psychology in the context of both K-12 mathematics learning and university Massive Online Open Courses (MOOCs). The experiments will be delivered through adaptive learning technology embedded in learning systems already being used by over 100,000 K-12 students and hundreds of thousands of MOOC learners each year. RAILKaM will also support 75 data scientists in conducting analyses on student data after the fact, using carefully redacted datasets that protect student privacy. In facilitating high-power, replicable experiments with diverse student populations and extensive measurement, this infrastructure increases the efficiency and ease of conducting high-quality educational research in online learning environments, bringing 21st-century research methods to education for the long-term betterment of learner outcomes.This project, RAILKaM, will support researchers in more easily running scaled, highly instrumented studies on education and educational psychology, both in K-12 and university Massive Online Open Courses (MOOCs). RAILKaM will leverage ASSISTments, an online learning platform for middle school mathematics homework and classwork used by more than 100,000 students each year. In addition, RAILKaM will build functionality atop the ASSISTments platform so that educational experiments involving scaffolded problem-solving can be easily built into MOOC courses. ASSISTments will use open source APIs to integrate with MOOCs offered by the University of Pennsylvania, branching capacity for investigation to higher education while enabling richer student interactions and data collection than is typically feasible in MOOC courses. These capacities will enable researchers to run online field experiments to test interventions designed to increase student learning and engagement with a focus on how adaptive learning experiences can be optimized. These experiments will be augmented by rich data collection on learners, extending MOOC log data and ASSISTments data with several indicators of learning and engagement not previously available for research at scale. This project will develop the software infrastructure necessary to conduct experiments and collect enriched data, as well as the social infrastructure necessary to select and refine study ideas while maintaining instructor control over the activities that students experience. The combined software and social infrastructure will enable us to engage with researchers who are interested in these issues but who currently lack the infrastructure, technical capacity, or access to learners necessary to conduct high-powered or complex randomized controlled trials. This infrastructure will help these researchers to improve scientific understanding of the principles of human learning, providing a unique shared resource for learning scientists that will have considerable potential for broader impact.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该项目RAILKaM将创建新技术，使20名研究人员能够在资助期间进行大规模的实地实验，在K-12数学学习和大学大规模在线开放课程（MOOC）的背景下研究教育和教育心理学的基本原理。这些实验将通过嵌入在学习系统中的自适应学习技术进行，每年已有超过10万名K-12学生和数十万MOOC学习者使用。RAILKaM还将支持75名数据科学家在事后对学生数据进行分析，使用经过仔细编辑的数据集来保护学生隐私。在促进高功率，可复制的实验与不同的学生群体和广泛的测量，这一基础设施提高了效率和方便进行高质量的教育研究在网上学习环境中，把21世纪的研究方法，教育的长期改善学习者的成果。这个项目，RAILKaM，将支持研究人员更容易运行规模，在K-12和大学大规模在线开放课程（MOOC）中，教育和教育心理学的高度工具化研究。RAILKaM将利用ASSISTments，这是一个用于中学数学作业和课堂作业的在线学习平台，每年有超过10万名学生使用。此外，RAILKaM将在ASSISTments平台上构建功能，以便将涉及脚手架问题解决的教育实验轻松构建到MOOC课程中。ASSISTments将使用开源API与宾夕法尼亚大学提供的MOOC集成，将调查能力分支到高等教育，同时实现比MOOC课程通常可行的更丰富的学生互动和数据收集。这些能力将使研究人员能够进行在线实地实验，以测试旨在提高学生学习和参与度的干预措施，重点是如何优化自适应学习体验。这些实验将通过收集关于学习者的丰富数据来增强，扩展MOOC日志数据和ASSISTments数据，其中包含以前无法用于大规模研究的几个学习和参与指标。该项目将开发必要的软件基础设施进行实验和收集丰富的数据，以及必要的社会基础设施，以选择和完善研究思路，同时保持教师对学生体验的活动的控制。软件和社会基础设施的结合将使我们能够与对这些问题感兴趣的研究人员接触，但他们目前缺乏进行高功率或复杂随机对照试验所需的基础设施，技术能力或学习者。这个基础设施将帮助这些研究人员提高对人类学习原理的科学理解，为学习型科学家提供一个独特的共享资源，这将具有相当大的潜力，产生更广泛的影响。这个奖项反映了NSF的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Toward Personalizing Students' Education with Crowdsourced Tutoring

通过众包辅导实现学生个性化教育

• DOI: 10.1145/3430895.3460130
• 发表时间: 2021
• 期刊: Lerning @ Scale 2021
• 作者: Prihar, Ethan;Patikorn, Thanaporn;Botelho, Anthony;Sales, Adam;Heffernan, Neil
• 通讯作者: Heffernan, Neil

Examining Student Effort on Help through Response Time Decomposition

通过响应时间分解检查学生对帮助的努力

• DOI: 10.1145/3448139.3448167
• 发表时间: 2021
• 期刊: Proceedings of the 11th International Conference on Learning Analytics and Knowledge
• 作者: Gurung, Ashish;Botelho, Anthony F.;Heffernan, Neil T.
• 通讯作者: Heffernan, Neil T.

Classifying Math Knowledge Components via Task-Adaptive Pre-Trained BERT.

通过任务自适应预训练 BERT 对数学知识成分进行分类。

• DOI: 10.1007/978-3-030-78292-4_33
• 发表时间: 2021
• 期刊: Artificial Intelligence in Education
• 作者: Shen, J.T.;Yamashita, M.;Prihar, E.;Heffernan, N.;Wu, X.;McGrew, S.;Lee, D.
• 通讯作者: Lee, D.

Effectiveness of Crowd-Sourcing On-Demand Assistance from Teachers in Online Learning Platforms

在线学习平台中教师众包按需协助的有效性

• DOI: 10.1145/3386527.3405912
• 发表时间: 2020
• 期刊: Proceedings of the Seventh ACM Conference on Learning @ Scale (L@S
• 作者: Patikorn, Thanaporn;Heffernan, Neil T.
• 通讯作者: Heffernan, Neil T.

Using Past Data to Warm Start Active Machine Learning: Does Context Matter?

使用过去的数据来热启动主动机器学习：上下文重要吗？

• DOI: 10.1145/3448139.3448154
• 发表时间: 2021
• 期刊: LAK21: 11th International Learning Analytics and Knowledge Conference
• 作者: Karumbaiah, Shamya;Lan, Andrew;Nagpal, Sachit;Baker, Ryan S.;Botelho, Anthony;Heffernan, Neil
• 通讯作者: Heffernan, Neil