Apologize if this is a naive question but I am new to this tool and want to double check. I am looking for the instructions/demos showing how to create the customized list of tags used in annotation. I am sure this is definitely doable for an annotation tool but can someone show me some pictures or docs describing that part of feature?
Hi! And yes, customising the label schemes is definitely a super important feature of an annotation tool and it’s definitely something you can do in Prodigy
What do you want to label? The annotation recipes typically let you pass in a --label
argument. For the manual recipes like ner.manual
, that’s the label scheme to use for annotation. For example, --label LABEL_ONE,LABEL_TWO
will let you annotate a text with two labels: LABEL_ONE
and LABEL_TWO
. Here’s an example command:
prodigy ner.manual your_dataset en_core_web_sm /path/to/your_data.jsonl --label LABEL_ONE,LABEL_TWO
You might also find the “Quickstart” sections at the bottom of the overview pages useful: see here for named entity recognition and here for text classification.