Text analytics approach to extract course improvement suggestions from students' feedback.

Area of Science:

• Educational Technology
• Natural Language Processing
• Data Mining

Background:

• Student feedback questionnaires are standard in academic settings.
• Qualitative comments within feedback contain valuable, yet difficult-to-extract, suggestions for course improvement.
• Manual extraction of suggestions from textual feedback is time-consuming.

Purpose of the Study:

• To develop and evaluate an automated solution for extracting explicit suggestions from student qualitative feedback.
• To enhance the process of gathering actionable insights for instructors to improve the learning experience.

Main Methods:

• The proposed solution employs text mining and data visualization techniques.
• It involves three stages: data pre-processing, explicit suggestion extraction, and visualization.
• Rule-based methods and statistical classifiers (including decision tree C5.0) were compared for extraction accuracy.

Main Results:

• The automated system successfully extracts explicit suggestions from student feedback comments.
• The decision tree (C5.0) classifier demonstrated superior performance in extracting and summarizing suggestions compared to other methods.
• The solution was validated using feedback from seven undergraduate courses.

Conclusions:

• Automated extraction of suggestions from student feedback is feasible and efficient.
• The decision tree (C5.0) model is highly effective for this specific text mining task.
• This approach provides a scalable method for academic institutions to leverage student feedback for continuous improvement.