Modeling link recommendations as a network growth mechanism and their impact on social contagion
TL;DR
It is shown that while simple contagions exhibit relatively modest shifts under most recommenders, complex contagions are highly sensitive to clustering-and homophily-based recommendations, thriving at moderate recommendation strengths but sharply diminishing under excessive recommendation strength.
Abstract
Link recommendation algorithms significantly shape online social networks, influencing both their structural evolution and critical processes such as information and behavior spread. This paper investigates how these algorithms affect simple and complex contagion processes by modeling recommendations as additional network growth mechanisms. We introduce a synthetic network model that integrates preferential attachment, triadic closure, and choice homophily, then extend it with various link recommenders, including heuristics and graph neural networks (GNNs). Our findings show that while simple contagions exhibit relatively modest shifts under most recommenders, complex contagions are highly sensitive to clustering-and homophily-based recommendations, thriving at moderate recommendation strengths but sharply diminishing under excessive recommendation strength. These results underscore the nuanced interplay between network structure, recommendation strength, and contagion dynamics, highlighting the importance of incorporating social contagions into the design of link recommendation algorithms.
