Tourist Place Reviews Sentiment Classification Using Machine Learning Techniques

Abstract

Social media is growing trend now a days. Every day millions of user review and rate tourist places on tourism websites. Sentiment analysis can be performed over these reviews which will be helpful to find tourist place popularity. Based on sentiment analysis result, tourist can easily decide tour destination to be visited. In this paper sentiment analysis has been implemented using machine learning approach. The Dataset has been collected from various tourism review websites. Here we have performed comparative study of feature extraction algorithms i.e. CountVectorization, TFIDFVectorization. Along with classification algorithms Naive Bayes (NB), Support Vector Machine (SVM) and Random Forest (RF). Performance of algorithms has been compared using various parameters like accuracy, recall, precision and f1-score. From experiment we found that TFIDFVectorization feature extraction algorithm has improved accuracy of classification algorithm as compare to CountVectorization for given review dataset. In sentiment classification of tourist place reviews TFIDFVectorization+RF has given highest accuracy 86% for a research dataset used.