jupyter json annotator

This package provides an annotation UI for arbitrary dataset in json format.

Install

pip install jupyter-annotator

Usage

1. Normal usage

from jupyter_annotator import Annotator

problems = [{
    "id": 2,
    "problem": "Where would I not want a fox? (a problem from coommonsenseQA)",
    "options": {
        "a": "hen house", "b": "england", "c": "mountains", "d": "english hunt", "e": "california"
    },
    "answer": "a",
    "filtered": "xxxxxxxxxx"
}]

anno = Annotator(problems)
anno.start()

2. Custom fields + skip + filter

• Custom fields: add custom field in the format (field_name, type, max_length)
• Skip fields: the fields which will not appear in the form but still in the preview so that they won't be edited.
• Filter fields: the fields that won't appear either in the form or in the preview

problems = [{
    "id": 2,
    "problem": "Where would I not want a fox? (a problem from coommonsenseQA)",
    "options": {
        "a": "hen house", "b": "england", "c": "mountains", "d": "english hunt", "e": "california"
    },
    "answer": "a",
    "filtered": "xxxxxxxxxx"
}]

custom_fields = [("rationale", str, 100)] 
skip_fields = ['id'] 
filter_fields = ["xxx"]

annotator = Annotator(problems, custom_fields=custom_fields, skip_fields=skip_fields, filter_fields=filter_fields)
annotator.start()

References

• Jupyter Widgets - Using Interact
• jupyter-innotater