Evaluation of online hotel reviews based on machine learning with sentiment analysis and topic modeling approaches

User-generated content in Web 2.0 platforms is a significant indicator of consumer satisfaction and service performance for hotel business. Revealing the structural and semantic attributes of this content plays an essential role in assessing the level of hotel service performance. By this way, hotels can precisely identify the factors influencing their consumers satisfaction. Machine learning techniques in the field of text mining effectively provides to analyze these contents within the scope of big data. In this study, online user-generated content related to hotels operating in Alanya, one of Turkey's prominent tourism centers, has been examined. The topics most emphasized by consumers who have stayed at these hotels and the effects of these topics on consumer satisfaction have been appeared via topic modeling and sentiment analysis methods. According to the findings, customers staying at the related hotels have identified animation, staff, and amenities as the topics that positively influence consumer satisfaction. However, front desk and room-related matters have been identified as the topics with the most negative impact. The staff topic has the most positive impact on consumers' positive comments but, the front desk topic tends to induce negative comments. In addition, the level of satisfaction with the topics revealed in the research varies according to the hotel class and travel types.