Efficient unveiling of multi-members in a social network

With the rapid growth of the Web 2.0, the discovery of key actors in social networks, called influencers, mediators, ambassadors or experts, has recently received a renewed of attention. In this article, we consider a particular type of actor that we call a multi-member since he belongs to several communities. We introduce a methodological framework to identify these actors in a hypergraph, in which the vertices are the actors and the hyperedges are the communities. We also show that detecting such multi-members is similar to the problem of the determination of a subset of minimal transversals of a hypergraph. An efficient algorithm that relies on the connection between the definition of a multi-member and that of an essential itemset is also introduced. Experiments done on several datasets showed that the introduced algorithm outperforms the pioneering ones of the literature.