The PI will undertake a three-year project to study opinion and content spreading in multichannel communication networks. Many people rely on online sources for their news and information; and people spread diverse content on these interconnected media. However, the content on media varies in quality and trustworthiness, from high-quality and thoroughly-researched content to puff pieces and misinformation. Both the quality and the viewpoint of online content, which the PI will incorporate into novel mathematical models of opinion and content spreading, have an enormous influence on online discussions on issues such as economic and social policy and on interactions with others of differing views. This project will study the structure of multichannel communication networks; analyze new mathematical models on these networks to gain a thorough understanding of the mechanisms of content spread interacting with opinion shifts of users on these networks; and use this analysis to develop strategies to flag content in a way that is analogous to spam filters, thereby helping guide users to better navigate content on various forms of media. The results of this project will also inform and feed back directly into theory and applications of networks in many disciplines, such as by improving methodology for studying how the structure of networks influences processes (content spreading, transportation, flow of nutrients, and others) that occur on networks. This project will support 2 graduate students each year.
The PI will undertake a three-year project to study opinion and content spreading in multichannel communication networks. This project has three components. First, using the new mathematical formalism of multilayer networks, the PI will analyze the structure of multichannel communication networks and develop network models of it in the form of multilayer random graphs. Second, the PI will generalize bounded-confidence models of opinion dynamics from monolayer networks to multilayer networks (this type of dynamical process has not been studied previously on multilayer networks, which one can use to encode networks with multiple types of edges or which consist of multiple subsystems), and the PI will systematically analyze the behavior of the these models through Monte Carlo simulations and regression analysis. Third, the PI will develop a model of interacting spreading of opinions (which are continuous-valued and can be in one or more dimensions) and content (which spreads through a discrete choice) by augmenting multilayer bounded-confidence models with media nodes that spread information of heterogeneous content qualities. Using this model, the PI will develop methodology to examine structural differences between the spreading properties of low-quality and high-quality content, thereby advancing the ability to filter disparate forms of content in networks. The results of this project will also inform and feed back directly into theory and applications of networks in many disciplines, such as by providing a principled method for determining the weights of interlayer edge in multilayer networks (a key open issue in applications of multilayer network analysis). The novel model of multilayer opinion dynamics will also influence the study of dynamical processes on networks more generally, advancing studies on novel phase transitions and bifurcations that arise from multilayer network architectures.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
