Feedback Analysis of Learning Evaluation Applications using Latent Dirichlet Allocation

The lecturer’s evaluation by a student (EDOM) application facilitates collecting anonymous feedback from students regarding their respective lecturers at the end of each semester. This feedback plays a vital role in evaluating teaching quality, identifying areas for improvement, and fostering a constructive learning environment. In this study, we propose leveraging the Latent Dirichlet Allocation (LDA) algorithm to analyze the substantial volume of student feedback data. The LDA algorithm offers an advanced solution for extracting topics from large document datasets, overcoming the challenges associated with laborious and time-consuming manual analysis. By applying the LDA algorithm to analyze 31,600 student feedback records, our experimental testing yielded a coherence score of 0.4, indicating the interpretability and coherence of the generated topics. The findings revealed significant topics, including students’ desire for enhanced learning experiences and suggestions for teaching methods. These insights offer valuable input for lecturers to improve their instructional practices and create a more engaging learning environment. Additionally, this study presents noteworthy experiments that determined the optimal topic modeling based on the number of topics, iterations, and relevance metrics.