\(p\)-means centrality (Q2206143)

This paper studies a generalized mean of the distances called \(p\)-means centrality in social networks. The proposed framework expresses the closeness centrality, harmonic centrality and eccentricity centrality using the generalized mean of the distances of all other nodes in the network to a given node. On the basis of a disease transmission SIR model, the paper evaluates the performance of the \(p\)-means centrality. The efficiency and feasibility of \(p\)-means are measured through numerical simulations with data from four real-world networks. The choice of the parameter \(p\) has also been discussed. It is shown that the \(p\)-means centrality provides better results for negative values of the parameter \(p\).