Replying to
@FactGrid@openbiblio.social Had a similar problem in Memarc for modern names in Eastern Europe. We used a mix of Geonames with fuzzy search (using Levenstein). Google geo api was also quite helpful, though in your case, it might be quite inaccurate. Still, you could try these, weeding out nonsense results, e.g. wrong continent. In your case, a phonetic similarity algorithm might help, on German you can use Kölner Phonetik.