Just wanted to clarify this one thing regarding Document Understanding x AI Center. If we want to deploy a custom ML package, we basically take up a labeling session, in which we define the custom field which is not actually present in the Out-of-the-Box package.
While doing so, we import the Out-of-the-Box package schema zip file, which eventually leads to the automatic incorporation of the column, regular and classification fields in the labeling environment. I noticed that this zip has a JSON file (which contains schema data) and a CSV file (which is apparently empty).
I wonder if is it possible to import the JSON which is being generated on creating a taxonomy? Since manually defining the taxonomy creates a JSON file too, I tried keeping it in a zip file & importing it in the labeling session, hoping that manually defined fields will be incorporated here as well, but I couldn’t figure it out.
What’s the catch here? Am I missing something? If it’s possible, what’s the procedure?
Maybe this part could be elaborated a bit more on what was actually done. We will be able to know the changes that you have made and if it is what was required to be done.
Sure thing. I took the Out-of-the-Box ML package schema available for invoices and imported the same during the data labelling session. As expected, all the fields defined in the schema reflected in the labelling.
As I noticed there was only a json and a csv file available inside this invoice schema zip folder, I wanted to achieve something simiiar with another json that I have. (This another json is being created as a result of taxonony creation).
My intention was to check whether importing taxonomy json can populate the fields in the labelling environment, as same the above mentioned scenario.
I checked both json side by side, but couldn’t find out the potential reason for which the importing wasn’t successful in the taxonomy json case.
Hope this elaboration is clear, please let me know if not.
Very precise. In fact, I compared both JSONs side by side but thought the taxonomy json might work as the initial import detected the type. Although, the structural difference seems to be the fundamental issue here.
I think I can directly make the necessary changes to the source json before importing it to the labeling environment, as you did with the name1 field.
