Taking China's movie theme parks as an example, the visitor review data of multiple tourism websites were collected, and the LDA theme model method, co-occurrence network analysis method and sentiment analysis method was used to study the tourist review text. The results show that the domestic movie theme park visitor reviews can be divided into four themes: tourism service theme, tourism landscape theme, overall perception theme and activity experience theme. Combined with the results of co-occurrence network analysis and sentiment analysis, the current situation and pain points of the prospect area are identified, which provides more accurate tourist preference information for cultural tourism enterprise operators, and also provides relevant theoretical and data support for the subsequent management and decision-making of enterprises.