Analysis of the Visual Presence of Political Party Leaders in Social Media (Bachelor Thesis, Finished)
The rise of social media has dramatically reshaped the way politicians interact with the voting public. As a consequence, the analysis of online communication of political actors has seen growing interest. However, existing research was mainly focused on textual content, despite the ubiquity of visuals on social media.
In this thesis, automated image analysis using standard computer vision tools is implemented and assessed. As a use case, face recognition and object detection are performed on more than 70’000 posts of Swiss and Canadian politicians. In the results thereof, differences and commonalities are explored based on attributes such as party membership and age. The posts were gathered from Facebook, Instagram and Twitter and allow us to also take into consideration cross-platform differences. Our results show that across all platforms, the politicians appear in more than 30 percent of their own posts and include party leaders in about 3 percent of their posts. It was also found that in all groups, the most frequently occurring object categories are the same and that noticeable differences can only be observed for objects that seldom appear.
Start / End Dates
2022/03/14 - 2022/07/13