E-Thesis 567 views 896 downloads
Lessons for Extremist Content Moderation: Capturing a Visual Style of Extreme Right Imagery / CONNOR REES
Swansea University Author: CONNOR REES
DOI (Published version): 10.23889/SUThesis.69262
Abstract
Online platforms have been subject to a recent surge in legislation from governments across the globe that mandate the swift detection and removal of harmful content (including extremist content) from their platforms, often under tight deadlines and with substantial fines for non-compliance. This tr...
| Published: |
Swansea University, Wales, UK
2025
|
|---|---|
| Institution: | Swansea University |
| Degree level: | Doctoral |
| Degree name: | Ph.D |
| Supervisor: | Nouri, L., and Müller, B. |
| URI: | https://cronfa.swan.ac.uk/Record/cronfa69262 |
| Abstract: |
Online platforms have been subject to a recent surge in legislation from governments across the globe that mandate the swift detection and removal of harmful content (including extremist content) from their platforms, often under tight deadlines and with substantial fines for non-compliance. This trend, which opposes the historical self-regulation by online platforms, stems from several factors, not least being the apparent connection between online extremist rhetoric and real-world violence. Identifying and eliminating extremist content poses significant challenges, particularly concerning material from the extreme right. The extreme right frequently operates on the fringes of tolerable communication and utilises tactics like dog whistles (hiding secret messages within content such as imagery). This research focuses on extreme right imagery due to how challenging it can be to moderate. Research that analyses the extreme right has historically received less scholarly attention compared to other forms of extremism, namely Islamist extremism. The study involves an analysis of a dataset of designated extreme right group images taken from the encrypted messenger platform Telegram. This dataset is used in a mixed methods approach to iteratively design an extreme right image analysis framework. The purpose of this framework is to characterise the unique qualities and identifiers of extreme right content. The results and findings from creating the framework and using it to assess over 25000 images used to inform data-driven policy recommendations for online platforms to enhance their moderation practices. In doing so, this thesis aims to contribute new knowledge to the field and aid in moderation against the extreme right whilst protecting the freedom of expression of other online platform service users. |
|---|---|
| Item Description: |
A selection of content is redacted or is partially redacted from this thesis to protect sensitive and personal information. |
| Keywords: |
Extreme Right, Online Extremism, Imagery |
| College: |
Faculty of Science and Engineering |
| Funders: |
EPSRC, Meta |