Embedding using custom data (PDF content along with images)

Hello,

I am building a chatbot using the custom data with embeddings approach. It works fine for a simple PDF document with textual data.

If my PDF file contains some graphics (like job aids, step-by-step process or an infographic), how do I handle the questions related to the textual data which are part of the graphics in the PDF file using embedding? Any suggestions will be very helpful. Thanks.

You could try programming an overlay to see where the preprinted portions land and script the margins to fill the parts you want… without seeing what the form I can’t write a template but it shouldn’t be too difficult