Detecting Visual Motives during the 2019 Swiss Federal Election (working-title) (Bachelor Thesis, Ongoing)


Yan Wang


Visual content has become increasingly important on social media. It is therefore not surprising, that images and visual motives play a crucial role in the dissemination of political messages and in influencing or even swaying public opinion. There exist many examples of visual themes that went viral in the course of a political discourse or during a major, political event. Since between 59 and 72% of people in the EU get their information on current events from social media, it is therefore desirable to have tools and techniques to identify and track such visual motives as they occur in the public discourse.

In this project, we are interested in analysing the spread of visual motives and memes in the course of the 2019 Swiss Federal Elections (“National- und Sta╠łnderatswahlen”) on Twitter. The goal is to leverage the near duplicate detection and clustering algorithms proposed in [1] to detect, track and ultimately visualize clusters of images that have been tweeted and re-tweeted by politicians, candidates or bystanders during said elections. The work in this project will be based on a dataset we have collected between July and October 2019. At this point it is unknown, whether that dataset contains such clusters of visual motives. However, it seems highly likely, considering discussions such as the one surrounding the “apple and worm”1 motive introduced by the right-wing Swiss People’s Party in the months up to the election.

As opposed to the work done in the original paper ([1]), we will try to build a system that works on online data. Hence, the system will receive a continuous stream of tweets in temporal order. The system is then required to (a) extract relevant features required for the given task (b) assign the image to an existing cluster or construct a new cluster from it and (c) visualize the evolving collection of images as they occur during a run.

[1] S. Zannettou, T. Caulfield, J. Blackburn, et al. 2018. On the Origins of Memes by Means of Fringe Web Communities. In Proceedings of the Internet Measurement Conference 2018 (IMC ’18). Association for Computing Machinery, New York, NY, USA, 188–202. DOI:

Start / End Dates

2020/02/24 - 2020/08/23


Research Topics