Enhancing the Value of Large-Enrollment Course Evaluation Data Using Sentiment Analysis

• Published in: Journal of chemical education Vol. 100; no. 10; pp. 4085 - 4091
• Main Authors: Hoar, Benjamin B., Ramachandran, Roshini, Levis-Fitzgerald, Marc, Sparck, Erin M., Wu, Ke, Liu, Chong
• Format: Journal Article
• Language: English
• Published: Easton American Chemical Society and Division of Chemical Education, Inc 10-10-2023; American Chemical Society
• Subjects: Attitudes; College students; Data mining; Feedback; Higher education; Latex; Learning; Natural language; Natural language processing; Sentiment analysis; Software development; Students; Teaching methods; Unstructured data; First-Year Undergraduate; Professional Development; Machine Learning; Student-Centered Learning; Administration Issues
• ISSN: 0021-9584; 1938-1328
• DOI: 10.1021/acs.jchemed.3c00258

In education, space exists for a tool that valorizes generic student course evaluation formats by organizing and recapitulating students’ views on the pedagogical practices to which they are exposed. Often, student opinions about a course are gathered using a general comment section that does not solicit feedback concerning specific course components. Herein, we show a novel approach to summarizing and organizing students’ opinions as a function of the language used in their course evaluations, specifically focusing on developing software that outputs actionable, specific feedback about course components in large-enrollment STEM contexts. Our approach augments existing course review formats, which rely heavily on unstructured text data, with a tool built from Python, LaTeX, and Google’s Natural Language API. The result is quantitative, summative sentiment analysis reports that have general and component-specific sections, aiming to address some of the challenges faced by educators when teaching large physical science courses.