Sentiment Analysis with Support Vector Machines for Travel Agency Ranking and Traveler Decision Support in Sri Lanka

Travelers often rely on online reviews to plan trips, but the large volume of data can hinder informed decision-making, especially in the Sri Lankan tourism industry. This research develops a sentiment analysis-based system to help travelers navigate reviews and make informed choices. The system uses Natural Language Processing (NLP) techniques to preprocess and structure user reviews, categorizing them into subcategories like transportation, accommodation, food and beverage, entertainment, recommendations, purpose, and travel agencys. Support Vector Machine (SVM) and Multinomial Naive Bayes (MultinomialNB) classifiers are employed to assess sentiment, classifying reviews as positive, neutral, or negative. The integration of NLP and SVM ensures high accuracy, efficiency, and interpretability in sentiment analysis, enhancing the model’s performance. The designed platform presents travelers with an easy-to-understand bar chart summarizing travel agency reviews and providing actionable insights into service quality. Key metrics like precision, recall, and score validate the model’s effectiveness. By organizing feedback into insights, the system empowers travelers to make better decisions and helps travel agencies identify areas for improvement. The scalable approach allows for adaptation to larger datasets and various tourism contexts, making it a reliable tool for comparing Sri Lankan travel agencies and improving the overall tourism experience.