The invention belongs to the technical field of tourism performance value evaluation, and relates to a tourism performance service value evaluation method, which comprises the steps of 1, obtaining influence factor indexes of tourism performance service values; 2, obtaining comment data and a comprehensive score; 3, obtaining a travel performance service value feature word list; 4, extracting feature sentences; 5, training an LSTM model; acquiring an emotion value; 6, calculating a gray correlation degree; 7, determining a topological structure of a gray neural network model; 8, training the gray neural network model; 9, obtaining the evaluation value of specific tourism performance play services. According to the invention, cultural elements are added to the evaluation of the tourism performance service value. An LSTM-based fine-grained sentiment analysis and gray neural network model is provided, so that the tourism performance service value evaluation method is enriched, and the evaluation is more intelligent and accurate. An index system is determined through gray correlation analysis, subjective and objective combination is achieved, and the method is more scientific.