Hello, I am very new to Prodigy and am looking to extend the labeling interface. Specifically, I am dealing with data with certain tags as metadata. E.g. multiple lines in jsonl can have the same file id in the metadata. There are certain triggers that would mean the whole file should be skipped and I would like to add a button for labelers to skip to the next file id based on the metadata (i.e. skip lines in jsonl until a different file id is encountered). Is this possible with Prodigy? Any help and suggestions are greatly appreciated.
I could be wrong, but I'm not aware of a way to get Prodigy to add an extra button that allows the user to send an extra signal to the backend to update the stream.
As a workaround, you might add some metadata such that an end-user can skip manually. Alternatively, you could also try to create a subset of your original
.jsonl file that only contains likely documents that should not be skipped. In my experience, you can get quite far by doing some heuristics in a Jupyter notebook before labelling.