Machine Learning Extractor OCR confidence has lower value


Based on my observation with the results of the ML - Receipts ( it shows a lower confidence level than the digitized document DOM.

Do the Data Extraction Scope and the ML extractor use the Digitized Document DOM or it has it’s own components inside?


Hi @Jeven_Delacruz

I believe @alexcabuz or @Ioana_Gligan should be able to tackle this one :slight_smile:

Hello @Jeven_Delacruz,

Currently the ML Extractor performs its own specific components inside - but this is work in progress to be changed.

On another note, please note that the confidence returned by the ML Extractor is NOT the OCR confidence, but the data extraction confidence. These two are different. The OCR confidence refers strictly to the confidence of recognizing the right letters, while the extraction confidence refers to the confidence that the reported value is from the right place.

1 Like

Hi @Ioana_Gligan,

I try to get knowledge in the below link. I have opened the sample project in the version 2019.7.0. But I am getting the below error . I have attached the manage package also here.

Erorr :

Manage and Packages:


Hi @balupad14

I don’t see this one in your error message, it’s our beta feed:

Most likely that then? (I’m not sure, just the first thought I had)


Hello @balupad14,

did you manage to get it working after you added the beta feed to the settings tab?

1 Like

Yes . It is working well. Once we got a solution from @loginerror. It will be not failed(just fun).


1 Like