Natural language processing
Text classification and measurement
Text analysis methods developed around a clear research question, checked against human coding, and documented for reuse.
Details →I am a computational social scientist based in Berlin. I work with research institutes, public agencies and civil society organisations. My projects use natural language processing to analyse text, surveys and experiments to study attitudes and behaviour, and platform data to examine questions of governance and participation.
Projects range from a single study to a work package within a larger programme or ongoing research advice.
An open-source tool for identifying political positions in text without task-specific training data. I led the development team, built and containerised the web application, and maintain its code repository.
A graduate course on using large language models in social research, with a focus on measurement. I also teach the material in one- and two-day workshops.
A peer-reviewed study examining what social media data can, and cannot, tell us about executive approval.
The first conversation about a possible project is free. If the work is outside my area, I will let you know and, where possible, recommend someone better placed to help.
juan@juangomezcruces.com