Understanding PDF and Bookmarking

General API discussion
#1

Hi,
I have the following requirement:

  1. Read an OCR file (Scanned bank forms in PDF format)
  2. Understand the document
  3. Get certain keywords and create a bookmark based on the keywords. For eg. I will have few keywords, the program needs to identify the keyword and bookmark on those pages the particular keyword appears. If I have keywords like alpha, gamma, upsilon, if the keyword Upsilon is found between page 1 and 3, it should create bookmark with name Upsilon. Similarly, if alpha word is found between page 4 and 10, those pages should be bookmarked with beta.
    If all the keywords are found it should create hierarchical bookmark based on the first found keyword, next and next.
    Is it possible to do with openai? If yes how?
#2

I think it could be possible but 1 it would be expensive and 2 I think gpt-3 is way overkill for this if it is just finding keywords. If you were to have gpt-3 detect which pages and parts of the pages could be helpful for a certain task, then I think gpt-3 could be a good fit.

1 Like
#3

You should look into writing code with GPT-3 that does this rather than using GPT-3 to do this.

1 Like
#4

How do you do it wiith GPT3?

#5

Why do you want to do it with GPT-3 so badly