With the advancement of modern UAV technology, UAVs have become integral to creating traffic management monitoring systems. Additionally, UAV-based traffic monitoring systems can predict traffic flow by integrating machine learning methods. Specifically, traffic flow data contains both spatial and temporal information, which can be effectively processed by graph neural networks (GNNs). However, GNNs often face the challenge of oversmoothing, which hinders their ability to capture complex structures in the data. The Spatio-Temporal Graph Ordinary Differential Equations (STGODE) model addresses this issue by introducing Neural Ordinary Differential Equations (NODEs) to construct deeper GNNs. Despite this, STGODE relies on initially predefined semantic neighborhood matrices, which do not adapt well to the dynamic nature of traffic information. To overcome this limitation, we propose an evolutionary graph neural network for traffic prediction, capable of continuously updating the semantic adjacency matrix throughout the training process. This dynamic evolution of the semantic adjacency matrix allows it to adapt to the features and semantic relations of the current data, enhancing its ability to capture the complexity and variability of traffic patterns. We validate our approach through experiments on several real-world datasets, demonstrating that our method outperforms state-of-the-art benchmarks.

Food recommendation systems are becoming increasingly vital in modern society, given the fast-paced lifestyle and diverse dietary habits. Existing research and implemented solutions often rely on user preferences and past behaviors for recommendations, which poses significant issues. Firstly, this approach inadequately considers the nutritional content of foods, potentially leading to recommendations that are overly homogeneous and lacking in diversity. Secondly, it may result in repetitive suggestions of the same types of foods, thereby encouraging users to develop unhealthy dietary habits that could adversely affect their overall health. To address this issue, we introduce a novel nutrition-related knowledge graph (NRKG) method based on graph convolutional networks (GCNs). This method not only enhances users’ ability to select appropriate foods but also encourages the development of healthy eating habits, thereby contributing to overall public health. The NRKG method comprises two key components: user nutrition-related food preferences and recipe nutrition components. The first component gathers nutritional information from recipes that users show interest in and synthesizes these data for user reference. The second component connects recipes with similar nutritional profiles, forming a complex heterogeneous graph structure. By learning from this graph, the NRKG method integrates user preferences with nutritional data, resulting in more accurate and personalized food recommendations. We evaluated the NRKG method against six baseline methods using real-world food datasets. In the 100% dataset, the five metrics exceeded the performance of the best baseline method by 2.8%, 5.9%, 1.5%, 9.7%, and 6.0%, respectively. The results indicate that our NRKG method significantly outperforms the baseline methods, including FeaStNet, DeepGCN, GraphSAGE, GAT, UniMP, and GATv2, demonstrating its superiority and effectiveness in promoting healthier and more diverse eating habits. Unlike these baseline methods, which primarily focus on hierarchical information propagation, our NRKG method offers a more comprehensive approach by integrating the nutritional information of recipes with user preferences.