How to use entity suggestions?

Start annotating manually. At least 5 or 10 annotations per label.

The system will observe what is happening and after a while start to train a first model using your annotation so far in order to calculate “suggestions”.

You may want to continue even after the first appearance of the blue “pop up“ anounces that suggestions have been computed (see What is a suggestion?).

After a decent amount of annotations (at least 10 for the simplest scenario),

  • Go to the Suggestions view as below:
Suggestions menu for NER project
  • Accept/reject/correct the suggested annotations (green check, red cross…)
Accept (green check) or reject (red cross) suggestions
  • Check that no other entities are present in the segment. If there is another entity which should be annotated, annotate it. 
Annotate missing elements in the segment
  • Validate the segment (or document). It will be added to the dataset with its annotation and be used in subsequent training runs.
Validate the segment to enrich the dataset
  • Sort suggestions according to their confidence level score.
Sorting suggestions
  • The suggestion engine is updated after few validations.
  • If the context of the segment is insufficient to validate a suggestion, increase the context
  • You can click on the title to access the whole document. Entities validation are only on the segment. The rest of the text is greyed.
  • You can filter (on the left) the list of suggestions on the tags you want to work on in particular!

Filtering suggestions
  • While browsing the suggestions and accepting / refusing them you will normally be able to proceed much quicker (generate more good examples & counter examples) than if you were to continue manually.

Let’s move on to the next step: How to review a dataset?