Sentiment analysis method for tracking touristics reviews in social media network

Abstract

The touristic sector in Tunisia has declined after the “Arabic Spring”. Therefore, the number of comments published by tourists to give their opinions about it has increased. Consequently, this resulted in a high volume of data in the different social networks such as Facebook and Twitter. In this case, the opinion mining plays an important role to more understanding and then ameliorating the situation of tourism in Tunisia. In this paper, the main goal is to select the tourists’ viewpoints in Twitter after the revolution. For this reason, we create a sentiment lexicon based on the emoticons and interjections as well as acronyms. We also use a sentiWordnet to build lexical scales for sentiment analysis of different tourist reviews with reference to a travel agency page on Facebook. Then, we propose a method relying on Support Vector Machine (SVM), Maximum entropy and Naive Bayes. Our approach is efficient as it gives encouraging results.