Measuring and Optimizing Influence for Resilient Community Networks

Abstract

Social networks and the use of technology allow communities to be connected, creating opportunities for individuals to spread information and influence others. This communication is critical when disruptions, such as natural disasters, occur. Finding these influencers, and subsequently maximizing their spread of influence in the network, is key for mitigating the effects of these disasters and restoring communities as quickly as possible. The proposed model seeks to first maximize the spread of influence through the network and then to minimize the vulnerability of the network after the disruption occurs. Maximization of influence involves a mixed integer formulation while minimizing vulnerability requires a bi-level function based on maximizing these influence scores before and after a disruption. The model incorporates social vulnerability scores to ensure the most susceptible members of the community are reached when needed. The network is subjected to disruptions by removing influencers of the community, affecting the most vulnerable members of the population, and creating spatial disruptions to disconnect the network. The model may be used to locate influencers and can be used by decision- makers to determine areas that need more assistance to be resistant to disasters. The model is tested on a sample graph with 16 nodes and applied to a Twitter network to find the influencers before and after a disruption.