Improving student success in subject domains that serve as gateway courses to STEM careers, such as chemistry and math, is of great interest. These subjects are difficult to teach and learn, partly because they involve a complex network of interconnected knowledge. Current online instructional materials generally follow a linear sequence, tracing a single predetermined pathway through the knowledge network. The CyberBook technology developed in this project instead guides students through the knowledge network along pathways chosen to optimize their learning of both the individual concepts and the relations between these concepts. The pathways adapt to individual students, with information gathered from a student’s interaction with the online materials to select a pathway that is optimal for that individual student. Although the primary focus is on chemistry and mathematics, the approach can be applied to other STEM domains. CyberBook combines traditional online courseware with intelligent tutoring systems. It also provides a research platform for learning scientists to conduct studies on how students learn. Beyond learning science, the proposed research has the potential to advance the literature on education (both domain-specific and domain-general), educational data mining, and artificial intelligence in education.
This project will develop several advanced learning-engineering methods to facilitate the creation of high-quality and effective online courseware: (1) an application of reinforcement learning to compute optimal sequencing of topics and scaffolding, (2) a text-mining method to automatically discover skills to be learned from text in written instructions and assessments, (3) a web browser-based method to rapidly create intelligent tutoring systems and seamlessly integrate them into online courseware, and (4) an application of reinforcement learning for an evidence-based quality control for online courseware content. As a proof of concept, instances of CyberBook for a college level chemistry (stoichiometry) and high school math (coordinate geometry) will be created. To validate the feasibility of implementation, the proposed learning engineering methods will be applied to two existing online course platforms—Open Learning Initiative and Open edX. The effectiveness of the proposed intervention will be measured through a pilot evaluation study at the partner institutions.
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.
