Sorry for the long waiting, I have checked with several product guys to figure this out since there is no document about this. The answer I final got is as below:
First we have a quick question to make sure there is no misunderstanding. Is the test document a different type or template than the one that the model was trained for or is it the same type but being given a low doc type confidence?
Answers to your questions below:
1) If DocTypeConfidence is high, it means the the tested document is likely the same type as the ones used to train the model
2) The tag confidence and doc type are separate, so in this case it means we are not confident that the test document is of the same type as the model but if they are the same doc type, we are confident about the predicted value
3) Depends on the situation - if the test document is different type the low DocTypeConfidence is expected, otherwise, are they able to extract the right values for their test doc?
For question 4, what do you mean by fixed element? If you use the same type of document (does this mean fixed element on your side?), it will do help the recogniton.
Regards,
Yutong