Hotel selection driven by online textual reviews: Applying a semantic partitioned sentiment dictionary and evidence theory

Browsing online reviews before selecting a satisfactory hotel has become a trend. Multiple criteria decision making models are powerful tools to provide competitive guidance. The first research gap motivating this study is that online textual reviews perform well in describing abundant perceptions and sentiments hidden in texts, while customer ratings used in existing models ignore them. Moreover, the existing sentiment analysis and hotel selection approaches have limited capacity in differentiating sentiment degrees, expressing natural languages, conveying comprehensive hotel descriptions and managing conflicting attitudes of different tourists. To narrow these gaps, a novel hotel selection model driven by online textual reviews on TripAdvisor.com is constructed. A semantic mapping function and the method of building this dictionary are proposed. Moreover, an evidence theory-based fusion method is proposed, which can guarantee the reliability of the results. Finally, the proposed model is tested in a case study and in robustness and comparative analyses.