Back to LanguageTool Homepage - Privacy - Imprint

404 for word2vec models :(

The documentation for the server states:

–word2vecModel a directory with word2vec data (optional), see https://github.com/languagetool-org/languagetool/blob/master/languagetool-standalone/CHANGES.md#word2vec

If I follow the link, I discover I have to download the necessary data:

The necessary data must be downloaded separately from https://fscs.hhu.de/languagetool/word2vec.tar.gz

But that link result in 404 error :frowning:

That code has been unmaintained for years, and it has never been active on languagetool.org. I think the only sensible step is to deprecate that option.

But it seems you can still get the files here: https://languagetool.org/download/word2vec/

  1. Is that in addition to the ngrams or in substitution of them?
  2. What’s the advantage over ngrams?

At the very least you have to remove the broken link from the docs.

It’s independent of ngrams. I cannot compare the two approaches, as I have never really used the word2vec approach.