Research on Online Shopping Mall Product Classification Recommendation Based on Graph Neural Network

The practical application of graph neural networks in online shopping malls is a significant area within real-life scenarios. As neural network technology continually advances, an increasing emphasis is being placed on how to efficiently leverage graph data for enhancing the functionality and performance of these e-commerce platforms. Nonetheless, prevailing methodologies commonly employ conventional adjacency matrices or basic attention mechanisms to consolidate and model data pertaining to items, users, and their interactions. To some extent, these approaches overlook the intricacy of graph structures and prove challenging to implement effectively on a grand scale. Running on graph data requires a large amount of computing resources, leading to performance issues. To address these issues, we propose a new method based on hierarchical transformation aggregation model to process graph data in online shopping malls more efficiently. This method first selects nodes at various levels and regions through random walks, and then uses graph convolutional neural networks (GCN) for information aggregation. Upon acquiring node features for each layer, a transformed attention mechanism is further introduced to more effectively consolidate the node features from every layer, thereby obtaining the ultimate representation of each node. This method better captures the local and global characteristics of graph data by considering node information at different levels, improving the performance and user experience of the mall. Ultimately, we substantiate the dependability of this model through multiple real-world dataset validations, thereby evidencing its practical application feasibility in online shopping malls. This method based on the hierarchical transformation aggregation model provides a more efficient way for malls to process and analyze graph data, and is expected to improve the performance of recommendation systems, personalized recommendations, and user shopping experience.