Analysis of adolescent social behavior on short video platforms based on graph neural networks

To accurately analyze adolescent social behavior on short video platforms, a Graph Neural Network (GNN) model is employed to construct social network graphs and extract features, enabling modeling and classification of interactions and behavioral patterns among users. The study explores the process of node embedding and behavior pattern aggregation and updating. The results indicate that GNN outperforms traditional machine learning methods in identifying behavior patterns and classification tasks, achieving an accuracy of 87.6%, particularly excelling in handling sparse data and complex social network structures. This research provides strong support for optimizing platform algorithms and enhancing user experience.