Research Question: This project analyzes the content of online political ads during the 2021 German federal election campaign to examine which political and social issues are salient across different localities and political parties and whether the policy platform of candidates matches policy demands of voters in a given constituency. 

Data: Text, speech, and video content of political ads on Facebook; local-level election results.

Methods: Webscraping; machine learning and natural language processing through Google Cloud Vision API and Video Intelligence and Speech-to-Text APIs

Challenges: Extracting content information from images and videos in political ads; rate limitations of Facebook’s Ad Library API; matching candidates’ political ads to municipal-level contextual information