DeepCURATER: Deep Learning for CoURse And Teaching Evaluation and Review

Zhihang Hu, Bhaavanaa Thumu, Yuren Qin, ZhiSheng Tao, Yijian Lu, Omar Kan, Yu Li, Irwin King

Abstract: Course and Teaching Evaluation (CTE) provides rel-evant and important feedback to a course that would help teach-ers identify potential issues in the curriculum, teaching materials, pedagogy, assessment, among others. The proper evaluation and review of the feedback surveys would lead to meaningful and ben-eficial outcomes for the stakeholders such as teachers, curriculum designers, school administrators, and ultimately learners as well. However, manually analyzing these surveys has many challenging issues, such as, objectivity, consistency, and scalability. To tackle these challenges, we present DeepCURATER, a CTE analysis system and its design, development, and implementation. The system collects unstructured survey texts and summarizes the output in the form of automatically-generated labels based on clustering algorithms by using sentence embeddings. The system also performs the aspect-based sentiment analysis of the survey with high precision. The DeepCURATER system is implemented as a user-friendly web-based system that allows users to explore the analysis results with visualization that would provide stakeholders with feedback to improve the teaching and learning environment.