چكيده به لاتين
The creative use of topological characteristics of nodes and the application of interaction, distribution and diffusion models have led to the emergence of various research achievements in identifying influential nodes in networks: most of the existing methods are based on considering the topological criteria of networks; Hence, the researchers compare each of the topological features to the quantities of famous models such as the gravity model with reference to interpretations, and then put their new model under test and draw conclusions and present it as a scientific research; In this way, there is still no standard for identifying influential nodes in complex networks.
The methods based on the use of gravity model in the identification of influential nodes provide a unified framework to evaluate the importance of nodes, but there are two bottlenecks that must be taken into account: (1) In most of these methods, the radius of influence of nodes is considered a fixed value. Which is obviously not true. (2) The location of the nodes is also an important factor to consider. The mutual attraction between each pair of nodes depends on their location, so that the nodes located in the central part of the network communicate with each other much more than the nodes located in the periphery. In this research, a basic action model between pairs of network nodes will be presented in order to calculate the influence score of the nodes, and also the short-range influence radius of the source nodes will be taken into account in order to be more consistent with actual behaviors. Allocating a significant contribution to the total impact score of a resource node due to the role of lower order neighbors compared to other neighbors is one of the goals of this research.
In this thesis, an action model for identifying influential nodes has been introduced, which is inspired by a physical model for determining the influence radius of nodes. This model, developed as Yukawa potential centrality (YPC), utilizes the concept of Yukawa potential and maps its parameters to the topological features of the network, enabling a dynamic and highly accurate evaluation of node influence. Unlike many existing models, YPC does not require assuming a fixed influence radius and calculates the influence radius of nodes based on the network's topological parameters. This capability, along with the model’s active nature, allows for a more precise and realistic analysis of complex networks.
The results of evaluations conducted on the data of real social and synthetic networks demonstrated that the YPC model successfully identifies key nodes with high accuracy and shows a significant correlation with the SIS benchmark model. Additionally, this model exhibits adaptability to various networks with diverse dynamics and structures. These achievements highlight YPC as an efficient tool for analyzing complex networks, making it applicable in various fields, including social, biological, and communication networks, providing deeper insights and facilitating better decision-making in these areas.
