SBIR Phase I: Solving the 2 Sigma Problem to Bridge the Academic Achievement Gap in Community Colleges

这个SBIR第一阶段的项目旨在通过建立一个网络平台来解决Benjamin Bloom的2-Sigma问题，在这个平台上，社区大学的教授和学生导师都可以远程帮助学生完成他们的课程。大多数社区大学的学生都是通勤学生，有工作和家庭需要照顾。他们不可能总是出现在校园里。该网络平台将利用校园现有资源、教授的办公时间和学生学习中心的可用导师，为学生提供远程帮助。这个项目的灵感来自布鲁姆在1984年的发现，他证明接受辅导的学生比98%的同龄人表现得更好。然而，这种制度在当时不可能对所有学生实施，因为它需要大幅增加大学的教育预算。因此，他将其称为2-Sigma问题，并预测未来的技术进步将使这种解决方案成为可能。对于全国的社区大学来说，这个提议的网络平台确实是这样一个解决方案。它可以满足国家对提高高等教育证书成就的关键需求，通过解决全国社区大学的低毕业率来帮助准备未来的劳动力。在美国，增加合格工人对于维持竞争力和经济增长是必不可少的。目前，两年制大学四年的毕业率不到20%，70%的入学学生没有拿到学位就离开了社区大学。这对美国的经济增长产生了负面影响，因为目前有700万个工作岗位空缺。损耗与整个学期积累的学习缺陷有关。拟议中的辅导平台将帮助学生在发现自己的学习缺陷后尽快赶上不太理解的概念，让他们能够直接向老师提问，并在为时已晚之前得到他们需要的即时帮助。技术上的挑战在于设计一个直观、用户友好的远程辅导平台，使其在没有任何培训的情况下就能在社区大学的教师和学生中获得较高的采用率。许多学生都有内存和处理能力有限的基本智能手机。该解决方案需要允许导师与学生进行远程交流，而不管他们的连接质量如何。它还必须让对技术有恐惧的教师为平台做出贡献，因为教师的参与对这种远程辅导解决方案的成功起着至关重要的作用。该技术还应易于与校园内所有现有的信息技术系统集成，并提供分析能力，以衡量其对学生成绩的影响。教授和学生导师将在网上列出模块，并安排可以同步或异步进行的辅导课程。我们将收集有价值的聊天/对话数据，然后在第二阶段，应用自然语言处理，将非结构化文本转换为结构化格式，然后应用机器学习算法构建一个可用于大多数学科的辅导人工智能助手。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值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 aims at solving Benjamin Bloom's 2-Sigma Problem by building a web platform where community college professors and student tutors alike can remotely assist students with their courses. Most community college students are commuter students with jobs and families to cater to. They can't always be present on campus. The web platform would leverage existing resources on campus, professors' office hours and student learning centers' available tutors, to provide remote help to students. The project is inspired by Bloom's findings in 1984 where he demonstrated that students receiving tutoring performed better than 98% of their peers. However, such a system was not possible to implement at that time for all students because it would have required significant increase in education budgets for colleges. Hence, he labeled it the 2-Sigma problem and predicted that future technological advances will enable such a solution. The proposed web platform is indeed such a solution for community colleges nationwide. It may serve a critical national need for increased post-secondary credentials achievement to help prepare the future workforce by addressing the low graduation rates in community colleges nationwide. Increasing credentialed workers in the United States is imperative to sustain competitiveness as well as economic growth. The current graduation rate stands at less than 20% after 4 years spent in 2-year colleges and 70% of enrolled students leave community college without a degree. This negatively impacts economic growth in the United States, as 7 million jobs are currently unfilled. Attrition is correlated to accumulated learning deficiencies that build up throughout a semester. The proposed tutoring platform would help students catch up on poorly understood concepts as soon as they identify their learning deficiency by giving them the ability to ask questions directly to their instructor and get the immediate help they need before it's too late. The technical challenge resides in designing an intuitive, user-friendly remote tutoring platform that can achieve a high adoption rate among instructors and students in community colleges without any training. Many students have basic smartphones with limited memory and processing power. The solution needs to allow tutors to remotely communicate with students regardless of the quality of their connectivity. It also must convert technophobe instructors to contribute to the platform, as instructors' participation plays a critical role in the success of such a remote tutoring solution. The technology should also easily integrate with all existing information technology systems on campus and provide analytical capabilities to measure its impact on students' grades. Professors and student tutors will list modules online and schedule tutoring sessions that can take place synchronously or asynchronously. We will collect valuable chat/conversation data and then during phase II, apply Natural Language Processing, convert unstructured text into a structured format, and then apply machine learning algorithms to build a tutoring AI assistant that can be used for most disciplines.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.