Kunqu Opera popularity heat prediction analysis based on Spatio-Temporal Graph Convolution Neural Network

Kunqu holds a significant place in the preservation list of the world's cultural heritage. By analyzing the spatial and temporal relevance in the development process of Kunqu, this paper proposes a Spatio-Temporal Graph Neural Network model (KQ-TGCN) that leverages historical performance information of Kunqu to predict its performance popularity. To further analyze the Kunqu performance popularity information, we propose Kunqu popularity and location density clustering algorithm. The spatial relevance is analyzed using Graph Convolutional Neural Network, and the temporal relevance is analyzed using LSTM. The experimental results indicate that the KQ-TGCN surpasses Graph Convolutional Neural Network (GCN) DCRNN, ST-Norm, TimeMixer and FC-LSTM methods in predicting the popularity of Kunqu performances and demonstrates a faster convergence during the training process.