OCR in only application

[Stanley Denman]

I am wanting to create an online application that takes a large pdf file and extracts information that is valuable for the user.  The key to the application is going to be speed - I am basically wanting to provide a minimal service for free that builds up an e-mail address. I know when I OCRed one of these files in FoxIt it takes about 20 minutes.  Here is my question: most of the information that I need is in the bookmarks but not all. One piece of info I need is an address that I could either get from accessing an API in Google Maps os something, or doing a partial OCR .  I can see OCRing 10-12 pages to get my info.   I am  wondering about speed - anyone have ideas about what approach would be the fastest?