unloc — a research application for a text corpus

We have a large collection of texts about the architecture of power in the post-communist bloc. We use an application to navigate interviews and workshop summaries, where LLMs work well.

We are expanding this knowledge with a very large literature corpus, in which we want highly specific searches — for example, by emotion. LLMs help, but there would be too much text to review everything on every query. Version 0 will therefore first select passages for further analysis.