Social-Behavioral Modeling for Complex Systems

19 An Artificial Intelligence/Machine Learning Perspective on Social Simulation: New Data and New Challenges

Osonde Osoba and Paul K. Davis

RAND Corporation and Pardee RAND Graduate School, Santa Monica, CA, 90401, USA

Objectives and Background

There is a growing demand to develop social and behavioral models competent to inform decision‐making in such diverse domains as counterinsurgency, political polarization, adversary propaganda campaigns, and public health behaviors. Success will depend on effective use of empirical information drawn from observation of both online and physical network human behavior. The objectives of this discussion are:

• To characterize the current infrastructure of data relevant to behavioral modeling.
• To describe progress on methods relevant to behavioral modeling that come from research on artificial intelligence (AI) and machine learning (ML).
• To identify shortcomings and challenges with current modeling approaches.
• To suggest where advances are needed to address them.
• To identify some mechanisms for doing so.

An earlier study conducted over three years by the National Research Council (Zacharias et al. 2008) presented a comprehensive review of social and behavioral modeling. Most of that analysis remains solid and apt today. Thus, we focus primarily on selected developments over the last 10 years.

Relevant Advances

Overview

Two trends relevant to this chapter have been notable over the last decade, (i) the burgeoning of data, ...