I have a requirement where User wants to extract any Pdf value whose format is not fixed. Means invoice or any other Pdf will be different ?. if yes then how can we build dynamic CDT value in IDP for document extraction of different extracted PDF.
Discussion posts and replies are publicly visible
Trying to extract "any" data is always a bit difficult. At the end the purpose of doing this is to get that data into a structure, and that structure needs to be pre-defined.
I suggest to try this with the new AI capabilities in Appian 23.2. It seems to be superior to the "old" IDP implementation.
Create the maximum number of columns in the tables that you can see near future can be utilized and train the IDP model using all possible files you having. In backend Appian uses google ML AI tool.
I just had a look at the new version, and you still have to predefine all fields.
But again, even machine learning is not very good in guessing and trying to translate random data into a structure.
Thanks for the detailed information stefan. It will really help.