Predicting and Influencing Individual and Group Attitudes and Behaviors with Gravity Models
Abstract
Modeling Political Polarization, Mobilization and Conflict De-Escalation throughGravity Models and Persuasive CampaignsApproved for Public ReleaseIn this proposal we aim to develop gravity models to understand online groups, their leaders andpropaganda themes acting as push and pull driving polarization, mobilization for collective actionand risky behaviors ??? such as joining extremist or anti-corruption groups, registering to participatea protest or a flash mob.Our design builds fundamental knowledge of social media tools that could be used to mitigate theinfluence of extremist groups and/or spread public affairs information regarding anti-corruptionand other development related initiatives. We also propose to develop new means of data gatheringand analysis that the intelligence community could employ to better understand the mindset andmotivations of non-state adversaries as well as potential civil-society partners.On social media information cascades driven by likes, shares, and comments can spark socialmovements. Social movements are about mobilizing people behind a shared purpose or goal. Theycan start out with just a small group of individuals who believe passionately in something and theycan end up changing the culture and politics around the world.However, what predicts what people like, share or comment on? In this proposal, our objective isto develop individual and group models for predicting and shaping social media behaviors such asliking, sharing, and more ???risky??? behaviors such as signing and sharing a petition, making acampaign donation, or registering to attend and share a protest event.We propose a four stage methodological framework to: (A) monitor individuals??? groupmembership, shifts in attitudes and transitions; (B) early-detect social dynamics such asfragmentation or coalescence; (C) apply gravity models to determine push and pull factors towardsand away from risky behaviors; (D) deploy such factors in targeted persuasive campaigns tofragment adversaries and coalesce allies. Our methods include social-psychological hypothesisand mathematical models that are closely integrated into campaign design such that we can predict,influence and ultimately mitigate risks posed by adversaries such as online extremist groups andstrengthen allied movements, such as counter-extremist and anti-hate movements.We will develop easy-to-use tool named, Persuasive Campaign Designer (PCD), with a graphicaluser interface that incorporate: (1) all methods presented in our testing environment, (2) all validpsychological concepts of persuasion. The tool will help an analyst to detect persuasive messagingcampaigns driving radicalization (i.e. positive feedback loops) and those that can be aimed fordeterrence of radicalization (i.e. negative feedback loops); that draw their content mainly fromcounter-extremist discourse segments within social media and assess their impact in Indonesianand Malaysian contexts. We will also identify effective symbols, contexts and persuasionstrategies in which ideological conflicts become magnets for hardliners, extremists and theiropponents. Similarly, we will employ the tool in Ukrainian and Georgian social media contexts todetect persuasive messaging campaigns incorporating groups, actors and propaganda themes inpositive feedback loops driving the development of anti-corruption activist groups.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 04, 2018
- Source ID
- N000141812761
Entities
People
- Hasan Davulcu
Organizations
- Arizona State University
- Office of Naval Research
- United States Navy