Online hotel reviews provide a vital information source for customers to select an optimal hotel, but a large amount of vague and unstructured information increases the difficulty of decision-making. From the perspective of customers’ risk attitudes, this paper proposes a novel fuzzy decision support model for hotel selection based on online reviews. Firstly, the useful information from online reviews is extracted by attribute extraction and sentiment analysis, and then this information is aggregated into the Probabilistic Linguistic Term Set (PLTS) by considering the weight of each review. Secondly, the improved linguistic scale functions are constructed from the perspective of customers’ risk attitude to convert PLTS into quantitative information. Thirdly, an integrated attribute weighting method is presented based on objective weights of the statistical measure and subjective weights of the Analytic Hierarchy Process (AHP) technique. Fourthly, an extended Combinative Distance-based Assessment (CODAS) method is developed to evaluate the performances of hotels. The effectiveness of the proposed model is verified by the practical case from TripAdvisor.com and the comparative analysis with the existing methods.