AI-Enabled Sentiment Mining of Homestay Reviews for Sustainable Tourism Quality Assessment in Uttarkashi

Abstract

A large number of user reviews have been generated via rapid growth of online travel platforms that provide rich information about visitor experiences and service quality. Unfortunately, the open-text nature of these reviews restricts analysis by manual means and is rudimentary in nature. The study presents an artificial intelligence-based framework for analyzing tourism quality assessment through hospitality reviews, owned by sustainable homestay tourism in Uttarkashi, Uttarakhand. The proposed framework is evaluated using a publicly available hotel-review dataset of 80,121 reviews. The methodology combines text preprocessing, sentiment classification and aspect-level analysis for visitor perceptions with respect to factors like service quality, cleanliness, location, room quality, value for money and sleep quality. A comparison of sentiment-classification performance between the conventional machine-learning models (Logistic Regression, Support Vector Machine, Random Forest and XGBoost) against a fine-tuned BERT model Then, the mined aspect-level sentiment patterns are interpreted to evaluate tourism-service quality and find suitable hotel strengths and weaknesses. The transformer-based BERT model outperforms traditional approaches for hospitality-review analysis, as evidenced by experimental results. Our proposed framework presents an intuitive and practical way to convert raw visitor feedback into tourism knowledge that is actionable, which can also help accelerate data-driven decision-making for sustainable tourism development and homestay quality improvement.