SBIR Phase I: Strengthening the Skills of the Parent With Online Coaching

这个SBIR第一阶段项目将开发一个机器学习辅助的家庭数学准备计划，帮助和鼓励父母为4-8岁的幼儿提供早期数学指导。与阅读一样，发展数学能力可以说是幼儿最重要的技能之一，特别是在STEM(科学、技术、工程和数学)在当今世界日益重要的情况下。然而，数学标准化考试的分数仍然低得惊人，只有40%的美国四年级学生的表现达到了年级的水平。现在大量的研究表明，这些差距通常在孩子第一次进入幼儿园之前就已经出现了。父母对孩子数学学习的强烈参与经常被认为是未来学业成功的关键指标，但实现这种参与的方法仍然难以捉摸。该项目以计算社会科学和建构主义学习模型为基础，旨在开发一种可扩展的方法，针对家长参与并促进家庭积极的数学学习体验，这些目标从根本上与NSF促进科学进步和促进国家福利的使命一致，并可能在未来几年产生重大的社会和经济影响。该项目将引入几项创新，特别是在人工智能和数据建模领域。将开发一款移动学习应用程序，为用户提供一系列屏幕上和屏幕外的数学活动，供父母和孩子一起参与。数据将从定期重复的短期调查中收集，这些数据将与应用程序提供的信息(学习数据、任务时间等)进行整理。将利用深度学习方法来探索内容、参与度和学习措施之间的关系，并建议预计将改善这些结果的后续活动。虽然数据集在第一阶段的研究中将受到限制，但随着参与程度随着时间的推移变得越来越强大，该技术将能够识别越来越多的连接。在第二阶段，这些工具和其他工具将描述和检测不止一个时间点的家庭订婚做法，不同的社会经济群体，以及各种不同的地理位置。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This SBIR Phase I project will develop a machine learning-assisted family math preparation program that will help and encourage parents to provide early math instruction to young children, ages 4-8. Along with reading, developing mathematical proficiency is arguably one of the most important skills for young children to acquire, particularly with the growing importance of STEM (science, technology, engineering, and math) in the world today. Yet standardized test scores in math remain alarmingly low, with just 40% of US fourth grade students performing at grade level. A considerable body of research now shows that these gaps often appear before a child first enters kindergarten. Strong parental engagement in a child's mathematics learning is often cited as a key indicator of future academic success, but ways to bring about this engagement remain elusive. Grounded both in computational social science and constructivist models of learning, this project aims to develop a scaleable means of targeting parental engagement and promoting positive math learning experiences for families, goals which are fundamentally aligned with NSF's mission to promote the progress of science and advance national welfare, and which could have significant social and economic impact for years to come. This project will introduce several innovations, particularly in the areas of artificial intelligence and data modeling. A mobile learning application will be developed that provides users with a range of on- and off-screen math activities for parent and child to engage in together. Data will be collected from short surveys repeated at regular intervals and these will be collated together with information provided by the app (learning data, time-on-task, etc.). Deep learning approaches will be leveraged to explore relationships between content, engagement, and learning measures and to suggest follow-on activities predicted to improve these outcomes. While the data set will be limited during Phase I research, the technology will be able to identify an increasing number of connections as participation becomes more and more robust over time. In Phase II, these and other instruments will describe and detect family engagement practices at more than just one point in time, with differing socio-economic groups, and in a variety of different geographies.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.