The Internet has provided many platforms for tourists to spread tourism-related information, resulting in a large amount of online review data on tourist attractions. Mining and analyzing online reviews by using modern information technology are of great importance. They affect tourists’ travel decisions and ensure the effective management of tourism attraction managers. We establish a multi-dimensional comprehensive evaluation indicator system based on the online reviews of 5A-level tourism attractions in 31 provinces and cities of China. We also utilize probabilistic linguistic term sets (PLTSs) to process the result of sentiment orientation and establish the integrated determination of objective criteria weights (IDOCRIW)-combined compromise solution (COCOSO) model to calculate the aggregate weight of attributes and rank the final evaluation of tourism attractions. Finally, we apply the proposed model to a case study on selecting tourism attractions and illustrate the effectiveness of this work.