Research on Ecotourism Recommendation Based on Graph Neural Network

Nowadays, the development speed of ecotourism is getting faster and faster, and with it, the phenomenon of ecotourism information overload has become increasingly prominent. In order to address the current problem of information overload in ecotourism, it is necessary to pay attention to the rational selection of ecotourism recommendation methods. Traditional ecological tourism recommendation algorithms mostly calculate similarity based on the basic attributes between tourists and ecological tourism projects, or from the evaluation of ecological tourism projects, and make recommendations on this basis. They largely ignore the influence of tourists’ own emotional factors and do not pay attention to tourists’ behavioral needs. Based on graph neural networks, this article extracts comment features from text, implements sentiment classification, calculates similar user groups, and finally evaluates the error of recommendation results through root mean square error, error, etc. The experimental results show that compared with the previous recommendation system methods based on Graphrec algorithm and Bia MF algorithm, the ecological tourism recommendation based on graph neural network can effectively reduce prediction errors.