Comparative Study on Key Characteristics of Shared Accommodation Based on LDA Model

The differences in perceptions of shared accommodation among tourists from different cultural backgrounds are yet to be explored. This paper uses 235,000 reviews of Airbnb accommodation experiences of Chinese and foreign visitors to Hong Kong as research content, and uses Jaccard similarity coefficients to calculate the co-occurrence intensity of high-frequency words and network clustering. The team experimented on the corpus using the Latent Dirichlet allocation (LDA) model, optimizing and adapting to two languages. In addition, this study also combines LDA thematic models with sentiment analysis methods. The co-occurrence network revolves around "room" and "landlord", while other words are fragmented, and the distribution of nodes in the co-occurrence network is more balanced among foreign tourists' experience elements. In terms of accommodation characteristics, both Chinese and foreign tourists focus on the host-guest interaction before staying, the location during the stay, the internal facilities and the feeling of home, while foreign tourists will have differences in paying attention to the city characteristics. The difference in the importance of Chinese and foreign preferences based on sentiment analysis is evident, with location being the most important concern for Chinese and foreign tourists, internal facilities being the most important concern and preferred by domestic tourists, while international tourists are more likely to identify with the 'home feeling' of shared accommodation, and the willingness to repurchase is not evident among international tourists, while Chinese tourists are more likely to recommend and repurchase. The comparative research idea of the experience dimension difference of Chinese and foreign tourists to Hong Kong in shared accommodation is proposed in this paper, and the conclusion can provide reference for the implementation of relevant measures by the government and small enterprises.