Finding Event-Specific Influencers in Dynamic Social Networks

Many methods have been proposed to determine influential nodes or links in network structures. However, most methods typically operate on a global scale with the assumption of static networks. Here a new ranking algorithm is proposed to find local influencers on Twitter that appear within the context of a specific event being discussed, incorporating the network dynamics as the event evolves with time. The algorithm is executed on data collected on Twitter during the Four-mile Canyon fire started on September 6th, 2010 in Boulder, Colorado. The ranking results are compared to other ranking techniques, and are shown to be the most effective in discovering local, event-specific influencers in the network during the fire.