<p>In information propagation networks, entity recognition faces challenges with traditional graph neural networks as they struggle to effectively capture both the static structure and dynamic evolution characteristics of networks. We propose a hybrid directed graph neural network approach that synchronizes static and dynamic tracks, consisting of two parallel and complementary learning modules: static features and dynamic features. The static feature module constructs a multi-dimensional structure perception mechanism by integrating intrinsic node features, the positional relationships of nodes in path sequences, and the length of path sequences. The dynamic module dynamically binds node features and position encoding to path sequences, using temporal neural networks to capture the temporal dynamic characteristics of information during propagation. The outputs of the two modules are fused through an adaptive graph attention mechanism to achieve joint representation of static structural information and dynamic evolution features. Experiments on multiple real datasets demonstrate that this method exhibits better performance in node recognition tasks. This research provides a new perspective for handling complex dynamic directed graphs, with potential applications in sentiment monitoring and network security.</p>