Global structure model modification to improve influential node detection

Improving a network's robustness and information acceleration requires assessing the value of its nodes, which has been a central issue in network research. The concept of centrality is crucial since it allows for determining the most important nodes. It is possible to find prominent nodes with the help of centrality indices, but they have computational complexity and are limited by the singularity function. The global structure model (GSM) is one method that helps find these impactful nodes. One of the problems with using GSM is that it ignores these nodes' local information. To address this issue, we propose that considering the features of each index individually and then combining them can result in more accurate detection of influential nodes. An experiment incorporated four attributes: global and local impacts, random walk structure, and node position. In this research, we simulate a real-world network using the SIRIR model to derive its propagation process and then verify its efficacy with measures like the Jaccard similarity score and Kendall's correlation coefficient. According to the findings of the experiments, the Degree of Centrality of the local features has a substantial effect when combined with GSM.