Stanza

Stanza is a Python NLP library for developers and researchers who need to analyze text in multiple human languages on their own hardware. It turns raw text into linguistic annotations, such as word forms, grammatical relationships, and named entities. It's open source under Apache 2.0.

Its pretrained neural models cover sentence splitting, tokenization, and expansion of words that contain multiple grammatical units. They also identify parts of speech and morphological features, recover base word forms, and parse dependencies between words. Named entity models identify names and other entities in supported languages; their language coverage differs from the broader parsing pipeline.

Stanza uses Universal Dependencies, a shared framework for grammatical annotation across languages. This makes it relevant to multilingual research and applications that need comparable linguistic output across different languages. Its native models run through PyTorch, and a GPU can speed up processing. Researchers can also train and evaluate the neural components with their own annotated data.

Biomedical and clinical English model packages provide syntactic analysis and entity recognition for research literature and clinical notes. For projects that need additional analysis, the Python interface to Stanford CoreNLP gives access to constituency parsing, coreference resolution, and linguistic pattern matching. That interface uses the Java CoreNLP package; the native Stanza pipeline runs independently of it.