These guidelines have been developed by ACE staff in conjunction with the ACE User Advisory Group and Internet Archive staff. They are intended for all books being digitized for the ACE portal, including those being processed at member libraries.
Markings and marginalia have been arranged into three different categories: red (markings which are vital to remove), yellow (markings which should be removed if possible), and green (markings which may be removed if staff judge them to be distracting).
Any mark that obstructs or obscured the text interferes greatly with the readability. The most critical of these are tight or messy underline that strikes through or touches the bottom of the characters. Other marks that interfere with multiple words or phrases, such as a large, messy circle, are also essential to remove.
Any underline or similar that is close to the text, even if it does not touch the characters, has the ability to interfere with OCR, especially for users with older versions of adaptive software (e.g. Kurzweill 1000).
Smaller markings such as asterisks, checkmarks, brackets, etc. within the text can add a few extra characters, but do not present a very serious threat to readability. If possible, such markings should be erased. If they interfere with an important part of the text such as a chapter title or citation information, it is even more important to remove them.
Marks that come directly before or after a line of text can be construed by some OCR software as extra characters in the text. If possible, it is preferable to erase marks such as slashes, asterisks, letters, or checkmarks that are directly adjacent to a character in the test.
Writing and symbols in the margins are, for the most part, easily ignored by most adaptive technologies. However, staff may choose to erase margin notes in certain cases based on personal judgement, if they wind them visually overwhelming or distracting.
Generally, highlighted text does not pose a problem for readability. The exception is particularly dark highlighters, or highlighting which is patchy and does not cover the whole character, which may pose problems with older software.
At the end of the day, it is a judgement call what to erase and what not to erase. Such judgements should be based on the knowledge of how markings affect the reading experience, as laid out above.