Context: Nowadays, people do not only navigate, but also contribute content to the Internet. Thoughts and opinions are written on rating sites, forums, social networks, blogs and other media. Such opinions constitute a valuable source of information for companies, governs and consumers, but it would be humanly impossible to analyze and locate the opinions in those assessments, due to the large volume and different origins of the data. For this, approaches and techniques of opinion mining in texts are used. Objective: To identify and characterize the techniques used for mining data in public opinion repositories regarding hotels, since the opinion mining area has offered necessary subsidies for decision-making related to hotel management. Besides, to identify, specifically, studies that investigated the opinions about the sustainability of hotels. Method: A systematic mapping was performed to characterize the research area. Results: It was identified that, among the main approaches, 31% of the works found use only data mining, while 55% exclusively use machine learning techniques, and 14% both. Conclusion: The most relevant studies in such research lines adopt machine learning algorithms such as Naive Bayes, SVM, LDA, decision tree, besides aspect-based techniques and SentiWordNet lexicon dictionaries. There are still opportunities to explore opinion mining solutions in online hotel reviews, mainly by taking into consideration aspects related to sustainable practices and sustainability levels practiced by each hotel.
展开▼
