spacy-stanza

Why we included this project

Teams that already standardize on spaCy but want the higher-accuracy Stanford models for linguistic annotation can get both without leaving their usual pipeline. The wrapper drops Stanza's tokenizer, lemmatizer, POS tagger, morphological analyzer, dependency parser, and NER into a normal spaCy pipeline, so results appear in the Doc and Token attributes you already use. Because the nlp object follows the standard spaCy Language API, existing tooling like displaCy, the rule-based matcher, custom components, and nlp.pipe for batched processing keeps working. You can also set the Stanza language separately, which helps for languages spaCy covers poorly. For multilingual text work, linguistic research, or NLP features inside larger applications, it is a practical way to combine two ecosystems.