Recently, techniques such as text mining and sentiment analysis have been applied widely to extract useful information from social network comments. to investigate Airbnb user preferences, this study adopts the Latent Dirichlet Allocation (LDA) model, and sentiment analysis to analyze the topic of 18067 users' reviews. Experimental results demonstrate that 12 topics including listing, host, location, experience features, and so on are the most important factors associated with user experience. Furthermore, more than 60% of the reviews express positive emotions.