Datasets:
europarl-de: build the German European Parliament reports
What I need from you: nothing.
European Parliament verbatim reports in German, CC-BY under the EP's reuse decision.
German is a full EP language, so this is clean, contemporary political speech — better than the OCR'd parliamentary material we already have. Port from the Dutch european_parliament or the Swedish europarl-sv implementation.
Order: after wiki-comments.
One correction before this is built: neither port in the plan works for German. Dutch european_parliament reads the Dutch-only release tvosch/GPT-NL-propella-annotations, and Swedish europarl-sv reads Språkbanken's Swedish europarl-sv.xml.bz2.
The EP Open Data API v2 (data.europarl.europa.eu/api/v2) is metadata only: /speeches has no text field, /speeches/{id} answers 500, and /documents/CRE-... carries no text. So the German reports have to come from the EP's published verbatim XML/PDF, not from the API. I will treat one document as one sitting report, as with europeana.