Abstract
The usage of social media is ever-increasing, and the number of users has reached 4 billion recently. Social platform has become an indispensable aspect of modern life and changed the way how people interact with one another. To assess individuals’ influence, social networks have routinely been developed to describe the relationships between individuals. In social network, everyone belongs to a community characterized by close internal connections and sparse external connections structure. Such structure is reflected in many real-world networks and has been widely investigated in the study of network science. However, the existence of community structure renders the network vulnerable to attacks and losses. No comprehensive community vulnerability evaluation model about the real-world network has been proposed to date. What type of community in formation should be considered, and how can they be considered in a com prehensive manner? Over-emphasis will lead to many duplication and invalid information, and inadequate consideration cannot reflect the structural connec tivity of the community properly. In order to fill this critical gap to determine the vulnerability of communities, we design a novel gravity-based model to evaluate the vulnerability of communities that considers multiple information. Specifically, (a) the number of edges inside the community, (b) the number of edges connected to neighboring communities, (c) and the gravity index are the three important factors used in this model for evaluating the community vul nerability. These three factors represent the interior information of the commu nity, small-scale interaction relationship, and large-scale interaction relation ship respectively. The gravity index of community is obtained from the fully connected undirected weighted community network which is constructed by iv the difference between communities. The fuzzy ranking of community is then developed to describe the relative vulnerability of each community compared to other communities. The sensitivity is further analyzed because weighting parameters are used to consider the contribution of each factor. We have validated and demonstrated the applicability of our proposed com munity vulnerability evaluation method via three real-world complex network test examples. The community vulnerability evaluation results from our pro posed model are expected to shed light on other properties of communities within social networks and have real-world applications across network sci ence. By evaluating these vulnerable communities, we can prioritize impor tant resources and personnel allocation to reduce potential damage, protect system performance, and improve emergency handling capacity, which is an indispensable step for many real-world multidisciplinary problems