Stanza NLP Library Documentation

Stanza supports Python 3.6 or later. Install the package using pip. This will also install dependencies like PyTorch (version 1.3.0 or above).

pip install stanza

You can train neural modules (tokenizer, MWT expander, POS/morphological tagger, lemmatizer, dependency parser, and NER) using your own data.

Data Formats:

• Tokenizer, MWT expander, POS/morphological tagger, lemmatizer, and dependency parser: CoNLL-U format.
• NER model: BIOES format.

Note: Model training is not supported via the Pipeline interface. You must clone the repository and run training from the source.

You can install Stanza using conda. Note: Anaconda installation currently does not support Python 3.10. For Python 3.10, use pip.

conda install -c stanfordnlp stanza

To develop on top of Stanza, clone the repository and install it in editable mode.

git clone https://github.com/stanfordnlp/stanza.git
cd stanza
pip install -e .

Stanza provides an official wrapper to access Java Stanford CoreNLP.

Setup Requirements:

1. Download Stanford CoreNLP and the required language models.
2. Place the model jars in the distribution folder.
3. Set the CORENLP_HOME environment variable to the path of your Stanford CoreNLP installation.

Example (nix): export CORENLP_HOME=/path/to/stanford-corenlp-4.5.3

To use the neural pipeline, import stanza, download the language models (e.g., 'en' for English), and initialize a stanza.Pipeline. You can then process text strings to create a Document object.

import stanza
stanza.download('en')       # Optional: pre-download English models
nlp = stanza.Pipeline('en') # Sets up a default neural pipeline in English
doc = nlp("Barack Obama was born in Hawaii. He was elected president in 2008.")
doc.sentences[0].print_dependencies()

To run the interactive demo locally, you must have stanza and flask installed.

1. Disable Browser CORS: Ensure your browser's CORS rules are disabled (e.g., using a Chrome extension).
2. Start the Server: From the stanza/pipeline/demo/ directory, set the FLASK_APP environment variable and run the Flask server:

   export FLASK_APP=demo_server.py
   flask run

3. Configure Client Address: In stanza-brat.js, uncomment the serverAddress declaration and set it to your Flask server's address (typically http://localhost:5000).
4. Launch: Open stanza-brat.html in your browser.

If you have an older version of Stanza installed, use the -U flag to upgrade.

pip install stanza -U

You can use a Streamlit-based web interface to visualize Semgrex and Ssurgeon results and process files. To set up the Streamlit webpage, follow these steps:

1. Install Streamlit: pip install streamlit.
2. Install Stanford CoreNLP if you haven't already.
3. Set the $CLASSPATH environment variable to point to your local Stanford CoreNLP installation.
4. Install the required dependencies: streamlit, spacy, and ipython. (Note: You can use the visualization stanza setup option to automate this).
5. Launch the application using the following command:

streamlit run stanza/utils/visualization/semgrex_app.py --theme.backgroundColor "#FFFFFF"

Stanza requires Python 3.6 or above. You can install it using pip.

!pip install stanza