I'd like to annotate text (aka. highlight and add a label) for the purpose of training NER (named entity recognition) models.
But it looks like none of the default templates can do this. And I've been doing a lot of googling but I haven't made much progress.
Ideally it might look something like below. Though I'm happy with anyway to achieve annotation.
I realize this is possible to do with Pybossa and Annotator.js but there doesn't appear to be any tutorials in existance.
Can anyone provide the steps/code to do this in Pybossa?
PYBOSSA can be used for this purpose without problems. All you have to do is load the text (sentences, paragraphs, ...) as tasks within a PYBOSSA project. Then, in the frontend you can use a library like this: http://annotatorjs.org/ and ask users to select the text that you want, annotate it and save it as a PYBOSSA task_run.
Then, you can use the webhooks solution to do the statistical analysis in real-time and train your model. Take a look at the webhooks documentation: https://docs.pybossa.com/build/webhooks/