<p>Online reviews rarely reveal which aspects of service quality drive sentiment and how these patterns differ across cultures, limiting their value for global operations. This study addresses this gap by examining how sentiments map onto SERVQUAL dimensions across cultures using bilingual Airbnb reviews. Leveraging a novel seed-guided classification and filtering approach (LDA and DFC) to isolate SERVQUAL-relevant texts, we compare dimension-level sentiment distributions. Results show that Western visitors place more (positive and negative) sentiment on service assurance, whereas Chinese visitors express more positive sentiments on service empathy and more negative sentiments on service tangibles. Accordingly, we develop culture-based service improvement strategies. Our study adds to the literature by demonstrating how to extract specific service knowledge from the sentiment distribution of interest and by showing that targeting sentiment to predefined service dimensions yields actionable insights for service operations. It also highlights the value of bilingual analysis in cross-cultural comparative research.</p>