Talk: Extraction of Party Positions from Political Texts

Title

Extraction of Party Positions from Political Texts

Abstract

Political debates can be analyzed at different levels of granularity, from fine grained identification of arguments to broad ideological distinctions. In the talk, I will present our work in the latter direction, aiming at the extraction of party positions using only embedding models and (mostly) no manual annotation. We base our studies on parties’ election manifestos. The structural meta-data that comes with these documents can be used to fine-tune SBERT representations for overall party similarity, and to an extent, for similarity at the level of policy domains. Our approach also generalizes to party position scaling across time and languages. Finally, we discuss the challenges of applying the approach to a more dynamic text type, namely social media posts.

Bio

Sebastian Padó is professor of computational linguistics at Stuttgart University since 2013. He studied in Saarbrücken and Edinburgh and was a postdoctoral scholar at Stanford University. His core research
concerns learning, representing, and processing semantic knowledge (broadly construed) from and in text. Examples include modeling linguistic phenomena (discourse structure, inference, etc.), applications in the computational social sciences and digital humanities, and methodological aspects such as model interpretation and robustness.

About the event

The talk will take place in the Auditorium Natural History Museum Denmark, next to the Pioneer Centre for AI. After the talk, there will be some limited time for one-on-ones with Professor Sebastian Padó (from 11-12), which you should sign up for when you register for the event.