home.social

#thoughtoftheday — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #thoughtoftheday, aggregated by home.social.

  1. #ThoughtOfTheDay: we invest significant efforts into reducing OCR error rates. At the same time we do not much to investigate the quality of our solutions for tokenization or lemmatization. Especially for genres that presumably were never seen in training (like e.g. historical forms of language). But we have benchmarks, you say. Yes, but again mostly for contemporary language, I'd guess. #dh #nlp #prodatphil