A Multimodal Approach to Network Information Dynamics
Abstract
Recently observed differences between the propagation of legitimate and misleading news items in social media have made it necessary to revisit traditional models of information dynamics over networks. The widespread notions of ÒcontagionÓ and ÒviralityÓ, in particular, while rightly putting emphasis on the interpersonal connections embodied by the networkÕs edges, fail to capture the complexities of the interactions between message and agent Ðin which human psycho-cognitive factors can play a central roleÐ as well as the intentional nature of the information emission and reception processes. In this project, we consider the interaction between incoming and outgoing messages and an agent as a dynamic process shaped by the agentÕs intents, beliefs and emotions, which can themselves be affected by the messagesÕ and the environmentÕs properties. In addition, the agentÕs motivation for information transmission, which determines her intent, is modeled as belonging to one of three modes: self-expression/ social signaling, monetary gain, and persuasion (also sometimes known as influence or manipulation). Our goal is to develop rigorous models of such multimodal information transmission which (1) explicitly account for intentionality while, (2) capturing the effects of psycho-cognitive factors such as bounded rationality, lack of bayesianity and influence of emotional states. These models are also (3) backed by Psychology and Communication Theory experiments and, (4) give rise to predictions that are testable either at the individual or network level. To this end, our team of control and game theorists, psychologists, communication theorist, economist, and computer scientist will take a multidisciplinary approach to address three interconnected research themes: (1) the development of models of intentional information transmission on networks, (2) their enrichment to incorporate behavioral elements such as lack of rationality and bayesianity and the role of emotions, and (3) their validation in experiments and on preexisting data sets. By establishing an understanding of the multimodal dynamics of information over networks, and studying how to predict and control these dynamics, this project will serve a critical need for future DoD and ARO operations and systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 09, 2020
- Source ID
- W911NF2010252
Entities
People
- Cedric Langbort
Organizations
- Army Contracting Command
- United States Army
- University of Illinois Urbana–Champaign