[Document Information Extraction] How to Efficiently Manage many Document Templates??

[minhpht294]

Currently, I have a requirement to build an BTP Application to read the Document from PDF file (PO, Invoice, SO, ...) then create a Draft version based on the data read from PDF file.

I found the Services: "Document Information Extraction" (DIE) and tested it with a simple PO. It is easy to create a template for a PO. However, I am thinking a scenario: "What if I have about 1000 BPs and each of them have a difference layout of PO? Do I need to create 1000 templates and train the DIE with 1000+ PO? How can I detect which PO go with which templates?"

I am new to this area and didn't find any document about this scenario. In the real world, how can I deal with that scenario?

Should I build a UI5 Application for BPs so they can upload their document and train the DIE instead of I do it all by myselfs?

[tobias_weller]

For processing PO documents, you can use our existing models for processing such documents by selecting the document type PURCHASE_ORDER and e.g. the SAP PO schema. That way, many formats might already be processed out of the box without the need to create templates (see https://developers.sap.com/tutorials/cp-aibus-dox-ui.html but use purchase order instead of invoice).

Another approach, especially in case you need to extract more information than our model supports, is to use our premium edition, this allows to extract most information without template creation (see https://community.sap.com/t5/technology-blogs-by-sap/introducing-document-information-extraction-pre...).