I would require some guidance on how to improve the question answering (language comprehension model) in Ui path AI center to answer questions from pdfs rather than just paragraphs. I am able to run the model in Ui path studio for answering questions regarding comprehension passages but required some information on how to extend its usage to PDFs as well.
I have tried using the extract document PDF text activity to extract text data from the pdf and then created a json formatted input ({“paragraph”: , “question”:}) for the ML skill but wasn’t able to apply the ML skill to the extracted data (paragraph) as it is showing errors in the inputted paragraph (the PDF which contains subheadings, bullet points, etc).
Thank you!