Online Review based Sentiment Classification on Bangladesh Airline Service using Supervised Learning
Khan Md. Hasib, Nurul Akter Towhid, Md. Golam Rabiul Alam
Abstract
Still leading main common modes of transportation we use on a daily basis is flying by airplane. Online reviews are becoming increasingly popular as a way for travelers to discuss their experiences with airplanes. Sentiment analysis may be used to understand people’s attitudes or sentiments by utilizing sites that provide opinion-rich data. To conduct this research, we generated a customized dataset including online reviews for 4 major Bangladesh Airlines, conducted a multiclass sentiment analysis, and compared the classifiers. This method begins with pre-processing procedures used to clean the reviews and balance the review data using the Pegasus model’s oversampling mechanism. System choices use various machine learning techniques to integrate feature engineering and word embedding for deep learning. The analysis was carried out 3 different machine learning (Decision Tree, Random Forest, XGBoost) and 3 different deep learning classification strategies (CNN, LSTM, BERT). Using a three-class dataset, the testing output is the online review sentiment (positive/negative/mixed) and the accuracy is determined. Based on the results, we have achieved the best accuracy 83% in terms of BERT. The accuracies were determined to compare each categorization technique, and the total sentiment count for all 4 airlines of Bangladesh have been displayed.