Machine Learning Form Analysis
After OCR recognition provide a facility not just for correcting the text, but for meta-data labeling of fields and regions. This data would make machine-learned document management much more effective.
For example, the automated extraction of data from Insurance company EOBs. You might argue, of course, that this is not in Adobe's balliwick, but I'd argue that this is a natural addition to your commenting tools.
Thanks for reporting your concern.
For the first question, ability to correct OCRed text. This functionality is already there. Go to Tools> Enhance scan> Recognize text> Correct Recognize text
For the 2nd query I will raise a feature request for you. We will look into this and inform you if there is nay update.