Harvester: Influence Optimization in Symmetric Interaction Networks

The problem of optimizing influence diffusion ina network has applications in areas such as marketing, diseasecontrol, social media analytics, and more. In all cases, an initial setof influencers are chosen so as to optimize influence propagation.While a lot of research has been devoted to the influencemaximization problem, most solutions proposed to date applyon directed networks, considering the undirected case to besolvable as a special case. In this paper, we propose a novelalgorithm, Harvester, that achieves results of higher quality thanthe state of the art on symmetric interaction networks, leveragingthe particular characteristics of such networks. Harvester isbased on the aggregation of instances of live-edge graphs, fromwhich we compute the influence potential of each node. Weshow that this technique can be applied for both influencemaximization under a known seed size and also for the dualproblem of seed minimization under a target influence spread.Our experimental study with real data sets demonstrates that:(a) Harvester outperforms the state-of-the-art method, IMM,in terms of both influence spread and seed size; and (b) itsvariant for the seed minimization problem yields good seed sizeestimates, reducing the number of required trial influence spreadestimations by a factor of two; and (c) it is scalable with growinggraph size and robust to variant edge influence probabilities.