Factors Influencing Airbnb Accommodation Prices in Busan Using Listing Information : Focusing on Keyword Analysis by Text Mining

This study examines the factors influencingper capita lodging prices using a hedonic pricing model and word cloud analysis techniques for Airbnb listings located in Busan, South Korea. Text-based keywords extracted from the listing descriptions along with the physical characteristics of the rooms were varied and their influence was compared by price range (low-end and high-end). The analysis showed that the number of bedrooms and bathrooms had a positive impact on price across all price points, while the presence of a private room had an opposite effect depending on the price point. In addition, host rating and number of reviews had a significant impact on price formation, and text-based keywords had different directional and explanatory power across price points. By integrating structured and unstructured data, this study provides a more sophisticated explanation of the pricing mechanisms on accommodation sharing platforms, and suggests the importance of establishing differentiated description strategies to meet consumer expectations. However, the analysis is limited to a specific region and time period, which constrains its generalizability, and the explanatory power for low-end listings was relatively weaker, indicating the need for additional variables. Moreover, the interpretation of keyword effects remains insufficient. Future research could extend the spatial and temporal scope of data collection and adopt complementary techniques such as sentiment analysis or topic modeling to provide a clearer understanding of how text-based descriptions influence price formation.