Sensing hotel customers distribution and their sentiment variations using online travel agent data: a case of Shanghai star-rated hotels

Abstract

Customer numbers and comments have been extensively studied to increase hotel arrivals and loyalty, whereas the specific distribution and sentiment variations have received less attention. This study constructed a data set of hotel customers and the deep learning model was applied to quantify unstructured text, which can mine the distribution and sentiment variation patterns of different star hotel customers. The findings show that customers of different star-rated hotels have different variation patterns of sentiment and distribution. There is a negative correlation between the visit heat and sentiment of some high-star hotels in various administrative regions: regions with lower visit heat have higher sentiment scores. In addition, season and location are important factors that affect customers’ sentiments. This study is the first to use a quantitative approach to examine the distribution and sentiment variation patterns of hotel customers. It provides insights into appropriate marketing strategies for the hotel industry and offers practical suggestions to hotel customers.

Keywords

• Customer sentiment
• Spatio-temporal patterns
• OTA data
• Star-rated hotels
• deep learning