Monitoring tourists' satisfaction has always been a concern for World Heritage Sites. Due to its ability to truly reflect tourists' travel experience and satisfaction evaluations, online review texts, after being mined and analyzed, can be promptly transformed into intuitive and highly valuable information, thus attracting increasing attention. This paper proposes a research method combining the Latent Dirichlet Allocation (LDA) topic model (hereinafter referred to as LDA) and tourist satisfaction responses, using user online reviews on Ctrip about the World Heritage Site of the West Lake Cultural Landscape of Hangzhou (hereinafter referred to as Hangzhou West Lake) and the Lijiang River Scenic Area in Lingchuan, Southern China Karst (hereinafter referred to as Guilin Lijiang River) as the original data. Ctrip provides two channels for tourists to evaluate their travel experience: satisfaction scoring and text comments. The Chinese comments are classified into themes through LDA in natural language processing, and the themes that tourists are most concerned about are identified as service, tourist attractions and travel experience. Through the 5-level satisfaction scoring (1 to 5) of tourists, a statistical analysis is conducted on tourist satisfaction and the themes they are most concerned about. The results show that the satisfaction under the service theme is the lowest, with nearly 11.5% of tourists at Hangzhou West Lake and nearly 21.1% of tourists at Guilin Lijiang River expressing dissatisfaction. Although there is a gap in tourist satisfaction between the two World Heritage Sites, both need to be significantly improved.