A dynamic propagation-based algorithm for node diffusion capacity evaluation in complex networks

Accurately evaluating the diffusion capacity of network nodes is crucial for understanding network structure and dynamic information propagation, with applications ranging from information dissemination and epidemic modeling to critical infrastructure protection. However, traditional centrality measures and diffusion metrics based on static network topology fail to capture dynamic propagation processes. To address this problem, we propose a novel dynamic propagation-based algorithm, termed NDC (Node Diffusion Capacity), for evaluating a node's diffusion ability. The proposed algorithm models propagation dynamics through three components-diffusion gain, loss, and stability-and quantifies each node's diffusion capacity by analyzing its propagation distance distribution and computing a Wasserstein distance. We conduct experiments on both weighted and unweighted networks, comparing NDC with 11 benchmark algorithms. Results show that NDC achieves a higher final infection ratio, shorter propagation duration, and greater average infection rate than baseline methods, with statistical significance confirmed by t-tests. Robustness tests under node removal, edge removal, and topological perturbations on Barab{\'a}si-Albert and Erd{\"o}s-R{\'e}nyi networks further demonstrate NDC's stability. Finally, experiments on 17 real-world networks (weighted and unweighted) indicate that NDC outperforms baseline algorithms in path survival rate, infection rate, redundancy, and propagation efficiency. These findings offer a novel perspective for assessing node diffusion capacity in complex networks.