Evaluating the service quality dimensions (SQDs) in the hotel sector is crucial to fostering competitiveness. To do so, it is necessary to analyse customer satisfaction, which can be measured through structured information from questionnaires or unstructured information from reviews. This research seeks to fill a gap in the literature by integrating a theoretical basis into the text mining of reviews and by analysing how customer reviews influence customer ratings. Thus, the theoretical foundations of the SERVQUAL model and its five dimensions of quality (tangibility, reliability, responsiveness, assurance and empathy) applied to five hotel service activities (rooms, restaurants, bars, leisure and hotel general services) are used to identify SQDs in the unstructured texts of hotel online reviews on TripAdvisor and to examine their relationship with customer ratings. For this purpose, the Bidirectional Encoder Representations from Transformers (BERT) model is used to automatically label the SQDs for each hotel service to determine which of these influence customer ratings. The resulting algorithm based on online reviews enables to easily determine which SQD of which service a hotel should focus its improvement efforts on in light of its customer satisfaction analysis.