Stanza — Stanford NLP Python Toolkit

Stanza

Stanza is Stanford NLP's production-grade Python NLP toolkit covering the full linguistic analysis pipeline. With pre-trained neural models for 70+ human languages, it delivers accurate tokenisation, multi-word token expansion, POS tagging, lemmatisation, NER, and dependency parsing in one consistent API.

Key Features

• 70+ language models, including low-resource languages
• Full NLP pipeline: tokenise → MWT → POS → lemma → depparse → NER → coref
• BiLSTM neural architecture; UD-trained dependency parsers
• spaCy-compatible wrapper (stanza.pipeline.core.Pipeline)
• Named entity recognition: 18+ entity types
• Biomedical/clinical NLP models (PubMed, MIMIC-III trained)
• CoreNLP Java server bridge for Stanford CoreNLP features

Quick Start

import stanza

stanza.download("en")  # download once
nlp = stanza.Pipeline(lang="en", processors="tokenize,mwt,pos,lemma,depparse,ner")

doc = nlp("Barack Obama was born in Hawaii. He was the 44th President.")
for sent in doc.sentences:
    for word in sent.words:
        print(f"{word.text:15s} POS={word.upos:6s} HEAD={sent.words[word.head-1].text if word.head > 0 else 'root'}")
    for ent in sent.ents:
        print(f"  NER: {ent.text} [{ent.type}]")

Install via ai-supply

npx ai-supply add stanza-stanford-nlp-toolkit

Curated mirror of the open-source Stanza (Apache-2.0). Get it from the source.