Would it be possible in Prodigy to allow users to select multiple text spans in a document and label each text span with a manually typed custom identifier? We have too many labels to show the user, but they are all integers and so are easy to manually enter. For example, it would be something like the ner.manual recipe, only instead of categories at the top it would be a text entry field.
Hi! In general, Prodigy tries to avoid free-form input wherever possible to speed up the process and ensure consistency – especially for label schemes. In most cases, you also want to be focusing on smaller numbers of labels at a time to reduce the cognitive load on the annotator and make it easier to run quicker and isolated experiments. So the workflow you describe currently isn’t easily possible – unless you build your own interface from scratch.