[ML Extractor ] Validation Station - View as Text

Help Document Understanding
1

Hi!

I was trying to process some sample receipts using the InvoicesReceipts Extractor - Demo workflow, and got rather curious about the ‘View as Text’ output in the Validation Station. Initially, I thought that the “View as Text” output would be reflected as the ML Extractor string output for the target fields (on the left side of the validation pane), but didn’t turn out that way.

Original receipt:

original
original1114×498 52.2 KB

“View as Text” Output

view_as_text
view_as_text1103×127 4.8 KB

MLExtractor Output

extractor_output
extractor_output765×591 26.1 KB

I also tried tweaking the OCR Engine in the ‘Digitize Document’ block (tried processing the same set of sample receipts with Microsoft and Tesseract OCRs), and sure enough the ‘View as Text’ output changed, but the ML Extractor output remained the same.

Digitize Document block

digitize_doc_block
digitize_doc_block512×645 30.5 KB

Anyone knows why? Thanks!

2

Did you ever resolve the mechanics between extractionResults and DocumentText input from the Digitze Document activity? I’m curious as I’ve finally had some time to start digging into this. Not sure if this is/was a bug that changed with later updates.