Hungarian model with spaCy

[oroszgy]

Hi,

Just letting you know that parsing Hungarian should be available through spaCy, as it can tokenize Hungarian and PoS tagging with Dependency parsing is also available through a 3rd party packages

Thanks,
Gyuri

[nikitakit]

Thanks for letting me know!

By any chance have you tried running benepar with your spaCy models for Hungarian? I'm always a bit nervous the spaCy integration introduced some sort of error that would be immediately obvious to a speaker of the language. My parser also has a POS tagger (optional but enabled by default), though I'm not sure how good it is compared to the spaCy model you linked.

I'll update the README with instructions on how to use Hungarian spaCy models. Sadly the tokenizer available via spacy.blank is not sufficient for benepar because it doesn't mark sentence boundaries.

[nikitakit]

Closing this now that I've added the link to the README. Thanks for telling me about the spaCy models (and let me know if you ever spot any issues with the Hungarian parser).