Date of Award

7-22-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Systems Engineering

First Advisor

Nitin Agarwal

Abstract

Focal Structures are key sets of individuals who may be responsible for online socially minded events, protests, or leading citizen engagement efforts on social media networks. Discovering focal structure sets who able to promote online social campaigns is important, but complex, as they are typically active locally, globally, and acting unlike central users and influencers, focal structure sets can influence large social media networks. Researchers applied a greedy algorithm to discover focal structure sets in social media networks. However, the outcomes were lacking on the users’ activities and interests that made the analysis incomplete. In this research, I present a bi-level decomposition optimization problem and a novel Contextual Focal Structure Analysis model (CFSA) to enhance the discovery and the interpretability of the focal structure sets to present influential sets of users and the contexts in terms of specific interaction patterns or communication structures. For the first time, the model utilized multiplex networks, where one layer could be the user network based on mentions replies, friends, followers etc. and the second layer could be the hashtag concurrence network. The two layers could have interconnections based on the user hashtag relations. The performance of the model was evaluated on three different real-world datasets including thousands of tweets related to “TrumpVaccine”,” BillGatesVaccine” and “Domestic Extremist Groups” and information related to COVID_19 pandemic, Black Lives Matter social movement and other related contents on Twitter in 2020-2021. The model discovered Contextual Focal Structures sets (CFS) associated with contexts in form of users-hashtags activities, revealed the contextual interests of users, and helped to answer, “what is going on between online users now?” Quantitative socio-technical methods such as deviant cyber flash mob (DCFM) detection and focal structure analysis (FSA) can provide reconnaissance capabilities that enable cities and governments to look beyond internal data and identify threats based on active events. Then I have validated the results based on three ground truth measures (modularity values, network stability, and average clustering coefficient values) to measure the influence of the CFS sets in the network, and finally, the Ranking Correlation Coefficient (RCC) was used to find the semi correlated solutions to be consistent with real-world scenarios.

Download

Share

COinS