Apply standard address to free form address data

Does anyone have experience developing a workflow to pull out address info from unstructured text and look up the corresponding, ‘proper’ value. I imagine machine learning would be a part of it.
I have a use case where invoices are coming from multiple sources but address information is not consistent. The result is that reporting is not rolling up invoice data by address as it should, with multiple entries for a given address. Address info is entered free form, which cannot be changed or controlled.

Thanks in advance.

I would say this is a poor use case for RPA. If data isn’t standard and cannot be controlled accuracy will always suffer.

Regardless, if you haven’t looked into OCR yet I recommend doing some research (Read PDF with Google Cloud OCR) (https://www.abbyy.com/en-us/solutions/content-iq-for-rpa/)