Date of Award

3-2022

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Operational Sciences

First Advisor

Phillip M. LaCasse, PhD

Abstract

Open Source Intelligence using social media is a practice which gives military intelligence analysts a window into the thoughts and minds of an online population. Using Social Network Analysis, user interactions on Twitter will be modeled as a weighted and directed network. Topic modeling through Latent Dirichlet Allocation uncovers the topics of discussion in Tweets and is then integrated into a multi-layer network which allows users to be connected to the conversations with which they have participated. Influential users in this network as well as highly connected groups of individuals are then discovered to paint a picture for intelligence analysts of the online landscape with which they are dealing. The results of this research demonstrate that the inclusion of topics in the social network allows for more robust findings in influential users when analysts collect Tweets from a variety of discussions through the use of more general search queries. PageRank was identified as the best performing influence ranking method for this problem context and two potential community identification methods were analyzed.

AFIT Designator

AFIT-ENS-MS-22-M-146

DTIC Accession Number

AD1172399

Download

Share

COinS