On Finding Small Sets that Influence Large Networks

TL;DR

This paper introduces a fast, simple algorithm for selecting minimal influencer sets in networks, outperforming existing methods in real-world scenarios and analyzing its theoretical and experimental performance.

Contribution

The paper presents a new algorithm for influence maximization that is both efficient and effective, with proven optimality in certain graph classes and superior real-world performance.

Findings

Algorithm outperforms existing methods in real networks.

Performance correlates with network modularity.

Propagation difficulty affects algorithm effectiveness.

Abstract

We consider the problem of selecting a minimum size subset of nodes in a network, that allows to activate all the nodes of the network. We present a fast and simple algorithm that, in real-life networks, produces solutions that outperform the ones obtained by using the best algorithms in the literature. We also investigate the theoretical performances of our algorithm and give proofs of optimality for some classes of graphs. From an experimental perspective, experiments also show that the performance of the algorithms correlates with the modularity of the analyzed network. Moreover, the more the influence among communities is hard to propagate, the less the performances of the algorithms differ. On the other hand, when the network allows some propagation of influence between different communities, the gap between the solutions returned by the proposed algorithm and by the previous algorithms in the literature increases.