I am working on a project where my goal is to extract data from a large diversity of pdf that does not follow a template (i.e unstructured data not template based). My ORC part works well and now I am trying to extract data from the raw text. During my research I have mainly seen data extraction from unstructured template based data.
The only things I see for now that could work, would be to use a regex, for example, to find a date, however, I don't really know what the date is related to, or to use NER, but I don't really see how to start.
I am not limited to NLP. I can also use computer vision on the pdf or other methods.
example of output:
{
date : "27/06/2019",
place : "saint-paul",
patient-name : Liliane marie SUZANNE"
patient-birthdate : "14/08/1957",
doctor : "METE",
type : "radiologie du thorax",
...
}
Any informations or solutions is welcome.