Online Social Networks (OSNs) usually exploit a logically centralized infrastructure which has several drawbacks including scalability, privacy, and dependence on a provider. In contrast to centralized OSNs, a Distributed Online Social Network helps to lower the cost of the provider drastically, and allows better control of user privacy. A distributed approach introduces new problems to address, as data availability or information diffusion, which require the definition of methods for the analysis of the social graph. This paper focuses the problem of the evaluation of the centrality of a node in a Distributed Online Social Network and proposes a distributed approach for the computation of the Ego Betweenness Centrality, which is an ego-centric method to approximate the Betweenness Centrality. We propose a set of algorithms to compute the betweenness centrality in static and dynamic graphs, which can be directed or undirected. We propose both a broadcast and a gossip protocol to compute the Ego Betweenness Centrality. A set of experimental results proving the effectiveness of our approach are presented. © 2014 IEEE.
Distributed protocols for Ego Betweenness Centrality computation in DOSNs
Conti M;Passarella A;
2014
Abstract
Online Social Networks (OSNs) usually exploit a logically centralized infrastructure which has several drawbacks including scalability, privacy, and dependence on a provider. In contrast to centralized OSNs, a Distributed Online Social Network helps to lower the cost of the provider drastically, and allows better control of user privacy. A distributed approach introduces new problems to address, as data availability or information diffusion, which require the definition of methods for the analysis of the social graph. This paper focuses the problem of the evaluation of the centrality of a node in a Distributed Online Social Network and proposes a distributed approach for the computation of the Ego Betweenness Centrality, which is an ego-centric method to approximate the Betweenness Centrality. We propose a set of algorithms to compute the betweenness centrality in static and dynamic graphs, which can be directed or undirected. We propose both a broadcast and a gossip protocol to compute the Ego Betweenness Centrality. A set of experimental results proving the effectiveness of our approach are presented. © 2014 IEEE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
