Principal Investigator: Gabriel Terejanu
CoPrincipal Investigator(s): Charles Pierce, Juan Caicedo
Organization: University of South Carolina at Columbia

Abstract:
This proposal is addressing the fundamental challenge of assessing individual student’s knowledge in cornerstone engineering classes with high student-to-faculty ratios. The goal is to: develop a computational assessment framework that easily integrates into an instructor’s routine efforts to track student knowledge, suggest remedial interventions, and guide future examinations. The rationale is that individual student knowledge is a hypothesis/model that needs to be tested using the scientific method. Similarly, assessment instruments are just experiments to discover how well a student masters specific concepts. This fits naturally with probabilistic methodologies such as the Bayesian inference that formalize the scientific method.

The main approach taken is to track the progress of individual students by developing student knowledge models based on Bayesian networks. This proposal addresses an open fundamental problem in constructing and using knowledge models to assess learning, namely how to relate curricular structure to knowledge models and how to inform the models using assessment data. The proposed methodology emphasizes the role of concept inventories to inform the construction of Bayesian networks models and to extract information from these models to suggest informative questions for future examinations. To facilitate ease of use and broader adoption by faculty, the software artifacts will be made available under open source licenses and the functionality of the framework will be integrated within a widely used learning management system through the development of a prototype plugin. The education and outreach aspects of this proposal include training participants in effective educational strategies and mentoring of future faculty.