Project Grant 2107008

This two-year, $299,647 Project Grant from the National Science Foundation's Engineering Education and Centers division will fund the development of human-in-the-loop natural language processing systems to analyze engineering education program feedback and assessment data. The awardee, Virginia Polytechnic Institute and State University, will build models in phases to cluster similar texts from multiple datasets on themes like students' systems thinking and open-ended course surveys. Researchers will evaluate consistency and refine configurations to handle different data types, like longer responses. They will also study how the systems impact human coder reliability, coding differences, time needed and differential treatment of student subgroups. All code and new datasets will be publicly shared. The work aims to augment human analysis of abundant program feedback text currently difficult to traditionally analyze at scale, with the goal of providing novel insights into engineering education ecosystems under the National Science Foundation's Engineering program (CFDA 47.041).