To see the problem:
- In an up-to-date installation of LIOS, open any PDF with multiple lines of text arranged into something OCR can't easily parse as a whole (for example, a table).
- Draw an area around each row.
- Finally, choose
Recognize -> Recognize-Selected-Areas and observe the results.
Notice that, in the text field below the PDF view, all the text from the various selected areas has been run together; this could be mitigated or improved by inserting a "\n" between the text from each area.