Case Study: Predicting Students Objectivity in Self-evaluation Responses Using Bert Single-Label and Multi-Label Fine-Tuned Deep-Learning Models

Students’ feedback data regarding teachers, courses, teaching tools, and methods represent valuable information for the education system. The obtained data can contribute in enhancing and improving the education system. Feedback from students is of great essence in the process of extracting hidden knowledge using various techniques for data mining and knowledge discovery. This paper presents various tools and methods for analyzing students’ feedback using Sentiment and Semantic analyses. The essential task in Sentiment analysis is to extract the particular sentiment from textual student responses in terms of negative and positive reactions, while the Semantic analysis contributes to combine textual response in the specific group based on questions. The output produced by the Sentiment and Semantic analyses provides a direct relationship between qualitative and quantitative parts of the evaluation in the form of student comments and grades.