Exploring the Use of Signs during Protest Activities through Social Media Data Integration: The Case of OccupyWallStreet
Date
Authors
Kash, Kathryn M
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Signs have long been used extensively in protest activities, such as political rallies, social unrest gatherings, picket lines, company boycotts, and marches. Consequently, recent studies have explored the use of posters in protests in terms of their role in the overall protest rhetoric, both visual and textual. While such studies articulate the central role signs have in protest activities, the locations of protest signs and their relationship to the spatial characteristics of protest activity often remain unexplored. Social media offer a new lens through which the location of signs in protest activities could be explored. In particular, the use of geotagged social media (such as Twitter and Flickr) contributions during protest activities can provide rich information about where signs are located, what narratives emerge from them, and how they are integrated into the overall activity. Toward this goal, this research proposes an approach for integrating geotagged Flickr images and Twitter messages for examining the relation between sign locations and protest activity locations. In this approach, the Stroke Width Transform and Optical Character Recognition are used to detect text in protest-related images. The location of these images is then compared to geotagged Twitter messages relating to the same protest activity, and patterns of interest are detected. The utility of this approach was examined through the analysis of the 2011 OccupyWallStreet protests across the United States. The results suggest that, overall, signs are immersed in the protest activity, but they tend to concentrate in specific locations that are likely to have a more central role in context of the protest.
Description
This thesis has been embargoed for 6 months. It will not be available until October 2016 at the earliest.
Keywords
Protest activity, Social media, Text detection, Contentious politics