SentiTaT: An Enhanced Approach for Tour and Travel Review Classification with Explainable AI

Sentiment analysis of Bangla tours and travel content is a growing area of research, particularly as user-generated content on platforms such as YouTube becomes increasingly prevalent. This study focuses on analyzing sentiments from 42,270 Bangla comments(Dataset link :https://github.com/kritanto1/Bangla-Tour---Travel-Dataset) collected from various travel vlog videos on YouTube. An extensive analysis was performed on 13 widely used machine learning (ML) and deep learning (DL) algorithms, utilizing four distinct feature extraction techniques individually. Among these models, the Artificial Neural Network (ANN) achieved the highest accuracy of 90% when paired with Word2Vec. To assess model performance, error metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) were calculated. Furthermore, Explainable AI (XAI) techniques, including Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP), were employed to enhance model interpretability and reveal the influence of features on predictions. The findings of this research provide significant insights for developing effective sentiment analysis systems in Bangla, particularly for tourism-related content.