Big data intelligent analysis modelling for predicting tourist behaviour in heritage sites

Cultural heritage sites, as invaluable carriers of human civilisation, attract large numbers of visitors.Accurate prediction of visitor behaviour is crucial for effective site management.However, existing research struggles to fully capture the complex dynamic changes in visitor behaviour, resulting in suboptimal prediction accuracy.To address these challenges, this paper first analyses visitor travel preferences based on improved term frequency-inverse document frequency and hierarchical clustering.Then, a spatio-temporal multi-scale graph is constructed to characterise the dynamic evolution of visitor behaviour across temporal and spatial dimensions.Next, graph neural networks are employed to extract and fuse features from multidimensional behavioural preference data.Finally, the transformer captures key spatio-temporal factors to achieve precise visitor behaviour prediction.Experimental results demonstrate that the proposed model achieves a weighted F1-score at least 4.14% higher than baseline models, providing scientific decision support for efficient heritage site management.