2010 | OriginalPaper | Buchkapitel
Discovery of Super-Mediators of Information Diffusion in Social Networks
verfasst von : Kazumi Saito, Masahiro Kimura, Kouzou Ohara, Hiroshi Motoda
Erschienen in: Discovery Science
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
We address the problem of discovering a different kind of influential nodes, which we call ”super-mediator”, i.e. those nodes which play an important role to pass the information to other nodes, and propose a method for discovering super-mediators from information diffusion samples without using a network structure. We divide the diffusion sequences in two groups (lower and upper), each assuming some probability distribution, find the best split by maximizing the likelihood, and rank the nodes in the upper sequences by the F-measure. We apply this measure to the information diffusion samples generated by two real networks, identify and rank the super-mediator nodes. We show that the high ranked super-mediators are also the high ranked influential nodes when the diffusion probability is large, i.e. the influential nodes also play a role of super-mediator for the other source nodes, and interestingly enough that when the high ranked super-mediators are different from the top ranked influential nodes, which is the case when the diffusion probability is small, those super-mediators become the high ranked influential nodes when the diffusion probability becomes larger. This finding will be useful to predict the influential nodes for the unexperienced spread of new information, e.g. spread of new acute contagion.