Uso de Inteligência Artificial para a Classificação de Comentários Turísticos do Jalapão nos Domínios da Experiência

Tourism is an essential activity for people’s well-being and for the economy surrounding touristic areas. When one is interested in improving and developing touristic activities, analyzing tourist experiences may provide useful insights. One way to understand tourist experiences is by collecting their comments on social media. This kind of analysis is made by interpreting people’s comments and classifying their experience as being related to entertainment, education/learning, evasion/escapism, or aesthetics/contemplation. These categories are called experience domains. This paper presents a solution for automatically classifying touristic comments in experience domains by using artificial intelligence (AI). To conduct the study, we applied a database with tourists’ comments about the Jalapão State Park (Parque Estadual do Jalapão, Brazil) that had been manually classified in experience domains. We tested different AI techniques, i.e., Gaussian Naive Bayes, Support Vector Machine (SVM), and LSTM, and demonstrated that the automatic analysis of comments is feasible. Our research method comprised five steps: 1) data preparation, 2) dataset partition, 3) training, 4) hyperparameters adjustment, and 5) prediction. Our results show that IA techniques present different performances, depending on how we prepare and partition data. The best results were obtained with SVM technique, with an 88% F1 Score.