Exploration and Prediction of Brand Communication Paths in Marketing Social Networks Using GCN

With the widespread application of social networks, accurate prediction of brand communication paths is crucial to corporate marketing strategies. Traditional brand communication path modeling methods usually only focus on one dimension (such as spatial or temporal dimension), while ignoring the spatiotemporal dynamics of brand communication. To this end, this paper proposes a brand communication path prediction method based on Graph Convolutional Network (GCN) and Variational Temporal Attention (VTA) mechanism. On this basis, this study proposes a VTA attention mechanism, which aims to improve the prediction accuracy of brand communication paths by refining the feature processing of spatial and temporal dimensions. First, this paper uses the GCN model to model the brand communication path, constructs elements such as brands, users, and advertisements into graph structure data, and aggregates the relationship information between nodes through the graph convolution layer to capture the spatial characteristics of brand communication. Then, in response to the dynamic changes of temporal information, this paper introduces the VTA attention mechanism and designs special attention modules for the spatial graph convolution part and the temporal convolution part to achieve refined weighting of the brand communication path. Experimental results show that the model using the VTA mechanism outperforms the traditional model in multiple brand communication path prediction tasks, with higher prediction accuracy and stronger generalization ability.