COMPETITION AND COLLABORATION IN COMPLEX NETWORKS

Abstract

Complex networks are crucial to the way we receive and process information; they make it possible for information and misinformation to spread with unprecedented ease. These considerations play out in settings across a range of different scales, from small teams interpreting data, to larger groups or global populations forming opinions based on the news they receive and share. Information in online environments is mediated both by social interaction and by algorithms that present the information. Our proposed work seeks to provide new insights on both the social and algorithmic dimensions of these questions. Our first goal is to develop new methods for analyzing socially mediated information. We will explore a set of models by which a group of agents linked in a network form beliefs based on information from their neighbors in the network and from globally visible external updates. We will investigate the power of interventions of different types, including adversarial interventions that seek to create discord, interventions that aim to protect the network against such adversaries, as well as competitive interventions in which multiple parties simultaneously seek to steer the distribution of beliefs. Our second goal is to substantially enrich the methods available for synthesizing and presenting information in online settings. One of the fundamental algorithmic primitives used in such settings is submodular maximization, in which we seek to select a set of items that optimizes a function. Submodular maximization has been shown to play a key role in many information applications from presenting recommendations to seeding information cascades. However, this framework is unable to handle the fact that in most applications, the items it selects have natural ordering or positional properties that affect the underlying objective. We will design new algorithms for submodular optimization in the presence of such ordering or positional constraints.

Document Details

Document Type
DoD Grant Award
Publication Date
Mar 06, 2024
Source ID
FA95502310410

Entities

People

  • Éva Tardos

Organizations

  • Air Force Office of Scientific Research
  • Cornell University
  • United States Air Force

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Theoretical Analysis.