A Deep-Learning Framework for Analysing Students' Review in Higher Education.

Area of Science:

• Educational Technology
• Natural Language Processing
• Machine Learning

Background:

• Tertiary institutions require student feedback for continuous improvement.
• Manual analysis of large volumes of qualitative student reviews is infeasible.
• Automated analysis of student feedback is crucial for efficient learning analytics.

Purpose of the Study:

• To present a framework for automated analysis of qualitative student reviews.
• To develop and evaluate models for aspect extraction, sentiment analysis, and grade prediction.
• To leverage student feedback for enhancing teaching and learning processes.

Main Methods:

• Aspect-term extraction using Bi-LSTM-CRF with BIO tagging.
• Aspect-category identification using Recurrent Neural Network (RNN) variants.
• Sentiment polarity determination using a Bi-GRU model.
• Grade prediction using a Bi-LSTM-ANN model combining textual and numerical features.

Main Results:

• Achieved a microaverage F1-score of 0.67 for aspect-term extraction.
• Developed 12 aspect categories for the education domain.
• Obtained a weighted F1-score of 0.96 for sentiment analysis using Bi-GRU.
• Predicted student grades with a weighted F1-score of 0.59, correctly identifying 20 out of 29 failing grades.

Conclusions:

• The proposed framework effectively automates the analysis of student qualitative reviews.
• The developed models demonstrate high accuracy in aspect extraction, sentiment analysis, and grade prediction.
• This approach offers a scalable solution for educational institutions to gain insights from student feedback.