PIs: Noboru Matsuda, Norman Bier, Larry Johnson
Texas A&M University
Most online courseware helps teach facts and concepts, while a different type of online learning software called intelligent tutoring systems can effectively teach skills in a way that is tailored to each learner. Unfortunately, these two tools are rarely integrated because of the expense and specialized expertise required to create intelligent tutors. This project will close this gap by building and testing a new scalable technology that will allow teachers without years of specialized training to author adaptive online courses that combine the best of both these approaches. This scalable cyberlearning platform will provide students with effective online instruction, provide learning engineers with an efficient authoring environment to build adaptive online courses, and provide researchers with a sharable corpus of big learning data that they can use to develop and refine theories of how students learn in adaptive online-course learning environments.
This project will build a web-browser-based authoring environment that supports the creation cognitive tutors and their seamless integration into online courses and will measure how well the resulting adaptive online courses promote facets of student learning such as synergetic competency and engagement. The central hypotheses are: (1) that the SimStudent technology — a machine-learning agent that learns cognitive skills from demonstration — can be a practical authoring tool for cognitive tutors that can be easily embedded into online courses; (2) that this technology can represent a tight connection between learners’ procedural competency and conceptual competency by combining knowledge-tracing (a standard method used by existing cognitive tutors) and text-mining (data-mining latent skills from traditional online course instructions) into an innovative student-modeling technique; and (3) that adaptive online courses created with this technique can produce robust student learning by promoting connections between their procedural and conceptual understanding (synergetic competency). As part of the overall research program, the project will: (a) develop a genetic application programming interface (API) for an existing web-based authoring technology to build cognitive tutors for online course integration; (b) develop an adaptive instructional technology as a generic control mechanism for adaptive online courses; (c) build new adaptive online courses on Open edX and also convert an existing OLI course into an adaptive online course; (d) conduct in-vivo studies using the adaptive online courses to test their effectiveness; (e) test the efficacy of the proposed adaptive online courses in supporting students to achieve the aforementioned synergetic competency. Successful completion of the project will yield the following expected outcomes: (i) a scalable online course architecture with efficient authoring tools for building cognitive tutors and integrating them into online courses in order to make those courses adaptive; (ii) a practical technique to identify relationships between procedural competency and conceptual competency; and (iii) an expanded theory of how students learn with the adaptive online course, and in particular of how students achieve robust learning with synergetic competency.