Recent PhD Jonathan Colner publishes on computer vision in Political Analysis
Quick Summary
- A new study by recent PhD graduate Jonathan Colner, now an assistant professor at American University, introduces a practical workflow for using free large language and object detection models to extract structured political data from poorly formatted sou
A new study in Political Analysis by recent PhD graduate Jonathan Colner, now an assistant professor at American University, introduces a practical workflow for using free large language and object detection models to extract structured political data from poorly formatted sources. While political scientists often rely on optical character recognition, regular expressions, or labor-intensive hand-coding, the method described in the article streamlines the process—requiring only a small number of hand-labeled examples to train effectively. Demonstrated on city council meeting minutes, the approach accurately isolates agenda items and could open the door to large-scale analysis of legislative, bureaucratic, and archival records in political science research.