Searching text in a PDF

Let me add to Eric's answer in that the results of OCR are problematic at best. I've been doing OCR for some 25–30 years and things have gotten better, but still often leave much to be desired. Things that can cause poor quality results include: very tiny text, ligatures (as mentioned by Eric), not clear text (photocopies are a good example), bleeding of the text on the other side of the page that was scanned, and some letter combinations often give issues such as "ui" being seen as "m." 

People have over-expectations of what should result including the rounding over of text when a page from a book is scanned. 

The reason why it "LOOKS" fine in the OCRed text is that Acrobat can remove the text from the page, put that on a different layer but the text you select is invisible. So it looks great but the actual text could be a mess. If you're looking for "compressors," that you can clearly see in front of you, it might very well be something altogether different in the OCRed text. 

One thing that can help is to get as good a copy from the beginning in the original scan. This can be a challenge because many scanners are now being released with scanning software that gives you no option to fine-tune the quality of the scan. Rather they give you options such as "Text," "Pictures," "Text and Pictures," and that's about it. I guess learning how to do a good scan was deemed too much of a challenge for companies. If you are fortunate to have reasonable quality scanning software with your scanner, this blog I wrote for Adobe sometime back may be of assistance.

http://photosbycoyne.com/Gary's_Help/Scanning/clean-scanning.html

Good luck!

You should be able to highlight the OCR'd text and see if some words were broken into seperate text frames. This happens sometimes when creating pdf or processing an OCR on a scan, saddly, we have no control were the 'text frames' start and stop. Sometimes it is a real mess.vanished, like ff link into a single character.

In the past, I encountered an issue were ligatures of the font generated characters that were not recognised or simply 

I never found a way around this.