Many large networks, such as social networks and web graphs, have been found to exhibit a variety of different interesting features. The dynamic properties of large networks, such as the propagation and robustness feature, have been actively investigated over the past years. In our research, we attempted to understand the propagation and robustness feature of large networks quantitatively. Specifically, we employed a critical parameter based on network topology, Edge Instability (EI), to study the propagation feature. Our experimental results indicate that, compared with the existing propagation strategies, the EI-based strategy has a more serious impact on information propagation in large networks. Furthermore, we investigated the robustness performance of large networks by evaluating their connectivity in the scenario that some edges are removed gradually. The experimental results show that, among the existing edge removing strategies, removing edges with high EI hurts network robustness most. Finally, in order to understand the influence of the average EI of a network, we devised an evolution model to generate large networks with varied average EI. We found that the average EI has a significant impact on information propagation. However, it has an unnoticeable influence on network robustness when EI-based strategies are adopted.

已经发现许多大型网络，例如社交网络和Web图形，都表现出各种不同的有趣功能。在过去的几年中，已经积极研究了大型网络的动态特性，例如传播和鲁棒性。在我们的研究中，我们试图定量地了解大型网络的传播和健壮性特征。具体而言，我们采用了基于网络拓扑的关键参数边缘不稳定性（EI）来研究传播特征。我们的实验结果表明，与现有的传播策略相比，EI基于策略的策略对大型网络中的信息传播有更严重的影响。此外，在某些边缘逐渐被删除的情况下，我们通过评估大型网络的连通性来研究其鲁棒性。实验结果表明，在现有的边缘去除策略中，去除具有较高EI的边缘对网络鲁棒性的损害最大。最后，为了了解网络平均EI的影响，我们设计了一种演化模型来生成具有不同平均EI的大型网络。我们发现平均EI对信息传播有重大影响。但是，当EI出现时，它对网络的健壮性影响不大采用基于策略的策略。