Hello,
I have several PDFs I need to extract information from.
For each PDF, there are multiple pages (5-8) but not all the pages are relevant, usually 3-5 pages are relevant.
For the pages that have the information I am interested in: it always has a user id that's not going to change and one value item that's like to change from relevant page to relevant page.
The catch is, however, the value item is not in a table format, rather like a line item, so I am not able to capture that with a table in my Document Form Definitions.
If there are 3 relevant pages in a 5 page PDF, my goal is to extra 3 sets of values (1 user id and 1 value item for each relevant page). However, if I am configuring my DFD to have only 2 line items and no table items, Decipher is only able to pick up the first set of values it can find on the first relevant page.
In order to combat that, I was told to use "burst pages into documents" to split a 5 pages PDF into 5 single-page PDFs. Howcwe, I have to manually right click on the thumbnail and make that selection. I have turned on the machine learning for the document type and trained it for at least 50 times, but Decipher is still not able to learn what to do in this situation, as a result - no matter how many times I upload the document to Decipher, it is still asking me to make human verification.
So my question is - Will decipher ever learn to use "Burst pages into documents" and if this is the best way to do in my situation.
Many thanks!
