Computational Ideal point estimation with textual data - a comparison of algorithms (Patparsch)

Description of the project

My PhD research investigates ideal point estimation of political parties and other political entities through computational methods using text data. This project compares a wide range of algorithms to assess their performance and efficiency in identifying political positions from textual sources. By systematically evaluating these algorithms, my research aims to uncover strengths and limitations, ultimately contributing to the improvement of computational tools in a range of social science disciplines like communication research, political science, or psychology.

My work not only enhances our understanding of political positioning but also advances methodologies for analyzing the vast, complex data associated with political discourse in a digital and algorithmically curated world. From an algorithmic or model perspective, my research therefore offers insights into how machine learning and other artificial intelligence systems (like large language models) produce and reproduce political biases in their output. You can find more information on my research on my website: https://patparsch.github.io/

Keywords

Ideal point estimation | Natural language processing | Algorithmic bias