MURI: SCAN: Socio-Cultural Attitudinal Networks

Abstract

The goal of the SCAN (Social Cultural Attitudinal Networks) project is to use the non-verbal content of multi-person group interactions in order to learn conditions that explain and predict several phenomena such as: (i) is a person in the group being deceptive? (ii) who is the most dominant person in the group? (iii) of two people in the group, who is more dominant? (iii) what is the level of trust that a given person in the group has in another given person? (iv) how much does a given person like another person? Do the answers to these questions vary by culture? Do the answers to these questions support existing theories of communication such as Dyadic Power Theory and the Topoi of Relational Communication? In order to answer these questions, the SCAN project has adapted a multi-player game called Resistance and an orchestration mechanism in which periodic surveys are interleaved into the game in order to gain ground truth, along with audio-visual recordings of the game. Next, the SCAN project aims to develop algorithms to extract diverse audio-visual features from the recordings of the game that may be useful in better understanding/predicting the phenomena of interest. Using this, SCAN hopes to define the concept of a social cultural attitudinal network in which players in the game are nodes and edges link players with information about their attitudes (e.g. level of trust, level of dominance). Using features from both the audio-visual content and the SCAN network, the project aims to develop sophisticated models capable of identifying deceptive behavior. Through a rich partnership between a collaborative team of computer scientists and social scientists, the SCAN project uses a mix of image processing, audio analysis, text processing, machine learning, and statistical methods to address these questions.

Document Details

Document Type
DoD Grant Award
Publication Date
Aug 06, 2019
Source ID
W911NF1610342

Entities

People

  • Larry Davis

Organizations

  • Army Contracting Command
  • United States Army
  • University of Maryland

Tags

Fields of Study

  • Computer science

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Game Theory.
  • Organizational Psychology.

Technology Areas

  • AI & ML