Information Diffusion Efficiency in Online Social Networks

The principle behind information diffusion is a key issue for online social networks, which has attracted much attention from both academia and industry. Some models, such as the independent cascade model, threshold model, and rumor model, have been proposed to characterize the process among nodes from the microscopic view. Topology properties such as degree and K-shell of the node have been used to evaluate the importance of a node. However, different topology properties will give different node's importance rankings in the process of information diffusion. The difficulty lies in the fact that previous work didn't consider the diffusion delay. In this paper, information diffusion is considered from a macroscopic perspective by taking time duration into account. A direct measure of node importance in information diffusion is proposed as a novel concept, diffusion efficiency. Based on the definition, a general information diffusion process is defined, where the aforementioned models can be directly generalized to their time-delayed versions. Simulation results show diffusion efficiency is better than degree and K-shell when measuring the importance of a node in the process of information diffusion. Furthermore, with taking time delay into consideration, the proposed information diffusion process becomes suitable for analyzing real online social networks.