Purpose: Few studies have applied sentiment analysis to identify service improvement priorities from Airbnb guest reviews. Therefore, this study aims to analyze guest review sentiments to generate insights for improving service quality and increasing reservations.

Methodology: A quantitative computational approach was applied to 599 Airbnb guest reviews collected via API scraping. Reviews were processed in Python, transformed using term frequency-inverse TF-IDF, and classified using the Support Vector Machine (SVM) algorithm. A word cloud was used to visualize the key review topics.

Results: The proposed SVM model achieved an accuracy of 98.33 %, demonstrating a robust sentiment classification performance. Most reviews (94.4%) expressed positive sentiments, highlighting property quality, host hospitality, and stay experience, whereas negative reviews (5.5%) primarily concerned cleanliness and facility maintenance. These findings identify key service attributes for improving guest satisfaction and supporting future reservations in the hospitality industry.

Conclusions: Guest perceptions are largely positive, indicating high satisfaction, although negative reviews remain important to be addressed. These findings demonstrate that sentiment analysis can identify service improvement priorities to enhance service quality and increase reservation numbers.

Limitations: This study focused on analyzing guest sentiment toward Tri Datu Properties on the Airbnb platform with 599 data entries.

Contributions: This study extends the Expectation Confirmation Theory by integrating sentiment analysis and machine learning to transform large-scale Airbnb reviews into actionable insights for hospitality decision-making.