SBIR-STTR Award

This STTR Phase I project proposes to develop and validate a student centered adaptive learning engine that is focused on improving learning outcomes using data collected from new and existing educational technology projects combined with advanced technology to automatically generate adaptive capabilities, thus creating ready-to-go intelligent tutoring systems. Providing adaptive instruction to students has been shown to be an effective way to improve student performance, yet very little educational software takes advantage of adaptive instruction due to high cost of creating adaptive content. This data-driven engine will significantly reduce the cost of adaptive learning by creating new methods of deriving intelligent tutoring capabilities from collected student data. Unlike pure machine learning solutions, this engine will allow for human input to maximize improvements through refinement over time. By using large datasets previously collected from existing tutors, these objectives can be tested and validated. The combination of human input with machine learning has the potential to make important gains in understanding student modeling. Finally, the engine will include new visualizations providing researchers, developers, and educators the tools to explore student data in ways that will allow for new insights into how students learn. The broader/commercial impact of an adaptive learning engine includes the ability to connect to educational software providing a service to software companies thereby, improving and extending their new and/or existing software to adapt to individual students and maximize learning. Adding adaptive instruction to existing software has traditionally been difficult due to the high costs of creating adaptive instruction. This engine reduces the cost of offering adaptive instruction capabilities by providing connections to existing and new software. Existing software can add capabilities without complete redevelopment, creating whole new markets for existing educational software companies, while bringing intelligent tutors mainstream. The Engine will provide educators new tools to understand how students learn with software systems. Key software providers in the K-12, Higher Ed, and Corporate/Government educational markets will enhance the learning of their students while maintaining existing training and tutoring tools. Companies and organizations are looking for effective online teaching and training solutions that are flexible to meet varying learning needs and preferences of users to maximize learning efficiency. This engine will connect existing expertise and research with the innovative vision to expand the capabilities of intelligent tutoring systems to reach to a variety of markets using a human-centered, data-driven approach.

This SBIR Phase II project represents a revolutionary advance in adaptive educational technology systems by using data collected from previous exercises to automatically generate hints and feedback for students. This work addresses the SBIR Educational Technologies and Applications subtopic EA5 - Learning and Assessment by making adaptive learning widely available and by providing tools to assess student performance in order to make interventions as early as possible and help students succeed. While it is well known that adaptive Computer Based Training (CBT) is more efficient and more effective, it has traditionally been cost prohibitive to produce in most domains. By leveraging the latest research in Big Data analytics and Educational Data Mining (EDM), this project will produce adaptive capabilities automatically, dramatically reducing the costs of producing more effective training and making such training widely available. The core customers for this technology will be providers of training systems. This includes publishing organizations, developers of software tools for education, and providers of corporate and government training. Institutions that are struggling to educate students, particularly across STEM (Science, Technology, Engineering, and Mathematics) fields, will be able to use this technology in their existing teaching systems and thus improve student engagement and performance.The final outcome of this project will be an integrated set of software tools that collect data from existing computer/web based training software, and automatically generate adaptive capabilities to create a personalized learning environment for students. It does this by using novel EDM and machine learning techniques to build and organize student and problem models that improve over time as more data is collected. It also provides for the tracking of student progress on specific concepts or skills (knowledge tracing), allowing for easy assessment at any point in time. The system also dynamically selects the students' next problems to maximize student learning and minimize time needed to master a set of skills (problem selection). For complex multi-step problems, this system will provide context-specific, just-in-time hints to help students as they learn. The final product will include data adapters that allow developers of existing software to seamlessly connect with this system. Finally, a main differentiator of this system is the transparent process of data curation and the related visualization tools that expose the problem- and student-model generation process. This combination of human input and machine learning will provide researchers, developers, and educators tools to explore student data and allow for new insights into how students learn.