The Impact of Passive Social Media Viewers in Influence Maximization

Abstract

A frequently studied problem in the context of digital marketing for online social networks is the influence maximization problem that seeks for an initial seed set of influencers to trigger an information propagation cascade (in terms of active message forwarders) of expected maximum impact. Previously studied problems typically neglect that the probability that individuals passively view content without forwarding it is much higher than the probability that they forward content. Considering passive viewing enables us to maximize more natural (social media) marketing metrics, including (a) the expected organic reach, (b) the expected number of total impressions, or (c) the expected patronage, all of which are investigated in this paper for the first time in the context of influence maximization. We propose mathematical models to maximize these objectives, whereby the model for variant (c) includes individual’s resistances and uses a multinomial logit model to model customer behavior. We also show that these models can be easily adapted to a competitive setting in which the seed set of a competitor is known. In a computational study based on network graphs from Twitter (now X) and from the literature, we show that one can increase the expected patronage, organic reach, and number of total impressions by 36% on average (and up to 13 times in particular cases) compared with seed sets obtained from the classical maximization of message-forwarding users. History: Accepted by Andrea Lodi, Area Editor for Design & Analysis of Algorithms—Discrete. Funding: This work was supported by the Federal Ministry of Education, Science and Research of Austria and by the Austrian Agency for International Mobility and Cooperation in Education, Science and Research [Reference ICM-2019-13384]. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2023.0047 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2023.0047 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .