Deep Learning-based Visitor Behavior Analysis and Prediction for Rural Tourism Intelligent Platforms

Abstract

With the continuous development of deep learning technology and the increasing maturity of rural tourism market, this paper obtains tourism user-generated content data through customized crawler technology, describes the data flow diagram of single-user crawling and the data flow diagram of database batch crawling module. A sentiment index covering multiple dimensions is constructed to mine the deep-seated features of tourist behavior. Fusing effective features in tourism data by using multiple topological maps, using graph convolution network to capture multiple spatial features of scenic spots and recurrent neural network to capture temporal features of traffic, to complete the analysis and prediction of tourists' behavior. Taking Jiangxi Wuyuan Huangling rural attraction market as an example for empirical analysis, the importance of historical flow and search volume under all time windows is as high as 111 and 117 respectively, proving that these two features have a significant impact on predicting the target variables. The model in this paper is highly fitted to the predicted value of actual passenger flow at 12 time points, especially in the 9th month, the predicted value is 402, which is 401 from the actual value, which is an important reference value for rural tourism management and marketing strategy.