Analysis of ISIS Twitter Media Content
Open Access
- Author:
- Fanelli, Michael
- Area of Honors:
- Data Sciences
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- Anna Cinzia Squicciarini, Thesis Supervisor
John Yen, Thesis Honors Advisor - Keywords:
- Classification
Machine Learning
Image-Recognition
Terrorism - Abstract:
- This thesis uses data from a newly collected source that is referenced here as the ISIS Twitter dataset and a Ground Truth web sample scraped from Google to create a learning model that can accurately identify the subject of an ISIS-related photo (being “Military”, “News”, or “Religion”). To begin, this thesis will introduce the motivation for this research was and provide a general overview of this thesis. Then it will address the datasets used as they are not in the public domain. Next previous research in the space of radical group media analysis will be discussed. An overview of the methodology for training and testing, as well as explanations of key concepts, will be addressed in a subsequent section. Results for the models used will be displayed and demonstrate that there is a capability to properly label images. Furthermore, conclusions from these results will be drawn. After this, limitations, and improvements for the research conducted will be addressed. Finally, future work that can be done with this thesis and a conclusion on the research will be discussed.