Identifying multiple influential spreaders based on generalized closeness centrality

• Published in: Physica A 2018-02, Vol.492, p.2237-2248
• Main Authors: Liu, Huan-Li, Ma, Chuang, Xiang, Bing-Bing, Tang, Ming, Zhang, Hai-Feng
• Format: Article
• Language: English
• Subjects: Complex networks; Generalized closeness centrality; K-means method; Multiple influential spreaders
• ISSN: 0378-4371; 1873-2119
• DOI: 10.1016/j.physa.2017.11.138

To maximize the spreading influence of multiple spreaders in complex networks, one important fact cannot be ignored: the multiple spreaders should be dispersively distributed in networks, which can effectively reduce the redundance of information spreading. For this purpose, we define a generalized closeness centrality (GCC) index by generalizing the closeness centrality index to a set of nodes. The problem converts to how to identify multiple spreaders such that an objective function has the minimal value. By comparing with the K-means clustering algorithm, we find that the optimization problem is very similar to the problem of minimizing the objective function in the K-means method. Therefore, how to find multiple nodes with the highest GCC value can be approximately solved by the K-means method. Two typical transmission dynamics—epidemic spreading process and rumor spreading process are implemented in real networks to verify the good performance of our proposed method. •An algorithm is proposed to identify multiple influential spreaders in complex networks.•A generalized closeness index (GCC) is given to maximize the distance among spreaders.•Finding the multiple nodes with the highest GCC can be approximately solved by K-means method.•The performance is validated by different spreading processes on different networks.