Analysis of Social Unrest Events using Spatio-Temporal Data Clustering and Agent-Based Modelling

Authors

Sudeep Basnet, University of Nebraska - LincolnFollow

First Advisor

Leen-Kiat Soh

Second Advisor

Ashok Samal

Date of this Version

8-2019

Citation

Basnet, Sudeep. (2019). Analysis of Social Unrest Events using Spatio-Temporal Data Clustering and Agent-Based Modelling (Master's thesis, University of Nebraska-Lincoln)

Comments

A THESIS Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Master of Science, Major: Computer Science, Under the Supervision of Professors Leen-Kiat Soh and Ashok Samal. Lincoln, Nebraska: August, 2019

Copyright 2019 Sudeep Basnet

Abstract

Social unrest such as appeals, protests, conflicts, fights and mass violence can result from a wide ranging of diverse factors making the analysis of causal relationships challenging, with high complexity and uncertainty. Unrest events can result in significant changes in a society ranging from new policies and regulations to regime change. Widespread unrest often arises through a process of feedback and cascading of a collection of past events over time, in regions that are close to each other. Understanding the dynamics of these social events and extrapolating their future growth will enable analysts to detect or forecast major societal events. The study and prediction of social unrest has primarily been done through case-studies and study of social media messaging using various natural language processing techniques. The grouping of related events is often done by subject matter experts that create profiles for countries or locations. We propose two approaches in understanding and modelling social unrest data: (1) spatio-temporal data clustering, and (2) agent-based modelling. We apply the clustering solution to real-world unrest events with socioeconomic and infrastructure factors. We also present a framework of an agent-based model where unrest events act as intelligent agents that continuously study their environment and perform actions. We run simulations of the agent-based model under varying conditions and evaluate the results in comparison to real-world data. Our results show the viability of our proposed solutions.

Adviser: Leen-Kiat Soh and Ashok Samal