Empower instructors with actionable insights: Mine and visualize student written feedback for instructors’ reflection

Student feedback on teaching at the end of the semester is an important source of information for instructors to gain insights into the effectiveness of their teaching. There are usually two forms of student feedback: quantitative scores and qualitative feedback. Quantitative scores can usually be easily summarized, while the analysis of qualitative feedback is usually effort-intensive as it deals with text. To help instructors glean insights from students' qualitative feedback, many previous studies used unsupervised approaches (i.e., topic modelling) for topic extraction in student feedback. Although topic modelling enables automated detection of previously unseen topics with minimal human effort, the generated topics are often incomprehensible and limited, as they were primarily derived from frequently occurring words. This study aims to extend previous research by developing a supervised text mining approach that integrates content analysis and a transformer-based pre-trained large language model to extract topic and sentiment categorization in student qualitative feedback. These categories are then visualized together with the quantitative scores to provide holistic insights for instructors' reflection and action. The purpose of this paper is to present the novel approach we developed to mine and visualize student qualitative feedback. It offers a holistic approach for higher education institutions to mine and visualize students’ quality feedback, providing instructors with actionable insights for improving their teaching practices.