ner-span-annotator 0.1.3

Streamlit component that allows you to visualize and modify NER annotations

NER Span Annotator

A Streamlit component for visualizing and modifying Named Entity Recognition (NER) annotations in the spaCy displacy span style. This tool is particularly useful for handling overlapping NER classes.

Key Features:

• Visualization: Displays NER annotations in a clear and intuitive span style.
• Interactivity: Allows you to modify NER annotations directly in the interface.

The implementation leverages code partially ported from Python to React, inspired by the spaCy repository. For more details about the spaCy span component, visit this page.

Installation instructions

pip install ner_span_annotator

Usage instructions

import streamlit as st

from ner_span_annotator import ner_span_annotator

spans = [
    {"start_token": 3, "end_token": 6, "label": "ORG"},
    {"start_token": 5, "end_token": 6, "label": "GPE"},
]
tokens = ["Welcome", "to", "the", "Bank", "of", "China", "."]

result = ner_span_annotator(tokens=tokens, spans=spans, labels=["ORG", "GPE"])

st.json(result)

Development

To set up and run the development environment, follow these steps:

Navigate to the ner_span_annotator/frontend directory and run:

npm install    # Initialize the project and install npm dependencies
npm run start  # Start the Webpack dev server

Then go to the project root and run the example app:

pip install -e . # install template as editable package
streamlit run ner_span_annotator/example.py # run the example