Can I approve/reject pre labelled text classifications

I have a rules based approach to extracting labels from text which relies on some ner and dependency parsing. I'm currently doing this in spacy and it assigns multiple labels to a piece of text.

I'd like to import these into Prodigy (db-in?) then have the annotator review these labels and approve or reject them (and possible update them) to improve on the rules generated labels.

Is it possible to do this in Prodigy?

Ideally, I would want the textcat.manual choice interface where a user could (1) accept (if all correct) or (2) reject or (3) correct and accept.

I have tried to import my rules based annotations (with db-in) then use review command to review them but I get an 'Oops something went wrong....' message when I load the Prodigy screen.

It works OK with classification view-id.

Using Chrome with Windows 10

Hi! You shouldn't have to import anything – importing is mostly for already annotated examples that you want to train from or combine with Prodigy annotations. The review recipe is mostly intended to review multiple overlapping annotations on the same data and reconcile them. If you want to label examples, you can just stream them in from a file, or using a custom recipe script.

In your case, it sounds like you probably want to stream in examples in the choice format with the label options and a list of "accept" IDs of the pre-selected categories – e.g. "accept": ["LABEL_ONE", "LABEL_TWO"]. This example of a custom recipe should be pretty close – only that you also want to add the task["accept"] populated by your custom rules.