Open MansMeg opened 10 months ago
For what it's worth I've cleaned up every headline of every motion from 1971 here in Wikidata: https://w.wiki/7u6o
You can compare it with the OCR errors in the Riksdagen scan from the same year and that might give some clues.
Thats very valuable!
And now I've done the same for 1972: https://w.wiki/84EP
Nice!
We want to predict which documents have poorer OCR errors to reOCR certain parts of the document. @MansMeg has submitted this as a data science project and hence have more detailed information.