NLP-Based Sentiment Analysis for Evaluating Student Feedback in English Language Education

Student feedback is essential responsibility of improving the quality of English language education. However, the evaluation of this form of feedback through manual analysis for training has for long been tiresome and seems to involve biased estimations hence a possibility of overlooking some key information. The objective of this research is the utilization of automated sentiment analysis with the help of LSTM networks to analyze students' feedback systematically and determine major issues and tendencies concerning improvement of the courses as well as general attitude towards the language classes. The contribution of this research is to use modern NLP tools to classify sentiments in educational feedback, which goes beyond the regular approaches to understand perceptions of students. The research framework presented covers every aspect of the methodology, such as data preprocessing techniques, feature selection through word embedding, use of LSTM network for model design, and performance evaluation. The proposed framework is useful in identifying the context and valuable to extract the overall sentiment when classifying performance feedbacks given by students. The proposed model demonstrates great performance in which the accuracy was at 98.5% with precision at 96%, recall at 90% and Fl-score of 91%. The category wise sentiment score for positive sentiments is quite high with teaching faculty (60%) and library facilities (75%) for negative sentiments it is quite low but we found out that course contents have been rated only 45% positively. Through analyzing the performance of LSTM for sentiment analysis of student feedback, the study demonstrates the importance of adopting sentiment analysis for evaluating feedback to help educators understand the areas in which learners need to be assisted to improve learning.