Examining Similar and Ideologically Correlated Imagery in Online Political Communication
share
This paper investigates the kinds of visual media US national politicians share on Twitter, how a politician's use of these various types of imagery reflects their political position, and identifies a hazard in using standard methods for image characterization in assessing political visual media. While past work has yielded useful results in investigating politicians' use of imagery in social media spaces, that work has focused primarily on photographic media, which may not be sufficient given the variety of visual media shared in such spaces (e.g., infographics, illustrations, or memes). This work instead uses clustering to characterize eight broad types of visual media US national politicians share in Twitter, where we find most politicians share a variety of types of imagery. Using three popular deep learning models to characterize imagery, results also show that how a politician distributes their imagery across these clusters is correlated with their overall ideological position. A qualitative assessment of these images reveals regularities among imagery used by liberal and conservative politicians, where conservative politicians tend to share patriotic imagery while liberals tend to share information-laden imagery. This analysis also reveals that all three image characterization models group images with vastly different semantic meaning into the same clusters, as confirmed in a post-hoc analysis of hateful memetic imagery. These results suggest that, while image-characterization techniques do identify general types of imagery that are correlated with political ideology, these methods miss critical semantic and politically relevant differences among images. Consequently, care should be taken when dealing with different varieties of imagery shared in online spaces.
READ FULL TEXT
