Structure-attribute similarity interplay in diffusion dynamics on social networks

Social interactions are shaped by homophily, the tendency for individuals to connect with others who share similar attributes. Exploring this phenomenon is crucial for understanding a wide spectrum of social behaviors, including the spread of misinformation and the dynamics of societal debates. In this study, we leverage a graph transformation strategy—which analyzes the interplay between individuals’ personal preferences and their structural connections—to investigate mechanisms of opinion/information diffusion. Among these latter ones, we focus on the Deffuant-Weisbuch model to simulate opinion dynamics and the Independent Cascade model to simulate information spread. Our findings on real-world social networks suggest that emphasizing attribute similarities enhances graph cohesion, whereas forcing structural similarities leads to fragmentation. Moreover, we observe a trend towards consensus opinion formation when enhancing attribute similarities, and faster as well as complete coverage of information spread in the same setup. These results motivate the importance of considering both individual attributes and network structure in studying social dynamics.