Comparative Analysis of Various Supervised Machine Learning Techniques Used for Sentiment Analysis on Tourism Reviews

Abstract

AbstractOn various social media sites, many tourists share their impressions and opinions in the form of text, photographs, and videos. To conduct sentiment analysis, these tourist reviews are collected and preprocessed. Text-based sentiment analysis currently relies on the creation of dictionaries and machine learning models to extract sentiments from text data. This form of sentiment analysis is commonly used in tourist satisfaction, tour recommendation, tourist place ranking, and decision-making. Traditional techniques in sentiment analysis in tourism domains are examined in this study. Using the Weka Tool, the experiment ran various machine learning algorithms on the tourist data collected from various websites and social media platforms. Lowercasing, lemmatization, stop word elimination, and nonword removal were all used as preprocessing techniques. The efficiency of classifiers was significantly improved as a result of these techniques. Sentiment processing was carried out using a supervised machine learning method in this experiment. This experiment's results are compared to one another. The outcomes are examined. As compared to J48, random forest, Logistic, and Naïve Bayes classifiers, LibSVM obtained the highest accuracy (92.5%) from the tourism data in this analysis.KeywordsSentiment analysisMachine learningOpinion mining