Here's an example (on OS X, they're stored in Documents > Digital Editions > Annotations):
<annotationSet xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xhtml="http://www.w3.org/1999/xhtml" xmlns="http://ns.adobe.com/digitaleditions/annotations">
<publication>
<dc:title>The Adventures of Sherlock Holmes</dc:title>
<dc:creator>Sir Conan Doyle</dc:creator>
<dc:publisher>Gutenberg Foundation</dc:publisher>
<dc:identifier>urn:uuid:c98702cc-1def-4789-9366-18d8bdf83710</dc:identifier>
</publication>
<annotation>
<dc:identifier>urn:uuid:0D5057C3-3E62-3522-DBE7-82686EA094E4</dc:identifier>
<dc:date>2008-01-16T11:42:37-08:00</dc:date>
<target>
<fragment start="chapter04.html#point(/1/4/4/14/1:3)" end="chapter04.html#point(/1/4/4/14/1:386)"/>
</target>
<content>
<dc:date>2008-01-16T11:42:37-08:00</dc:date>
<text>example of an annotation</text>
</content>
</annotation>
</annotationSet>
The 4/14/1:3 and 4/14/1:386 say that the annotation stretches from the 3rd to the 386th character of the 14th node (which is actually the 7th paragraph), I believe.
However, what if you want to store the annotations within the XML document itself? With the possibility of multiple overlapping annotations and alternatives for a certain section of text, this is the best I can think of at the moment:
<doc>
Boscombe Valley is a country district not very far from <annotation length="4" content="a town"/>Ross, in Herefordshire.
The largest landed proprietor in that part is a <annotation length="15" content="a person"/>Mr. John Turner, who made his money in <annotation length="9" "a country"/>Australia and returned some years ago to the old country.
</doc>
This uses a closed annotation element to mark the start of the annotation, and a length attribute to denote the end of the annotation. The content of the annotation is stored in an attribute, and other attributes can be added as necessary in the same way.
If anyone knows any other prior art, I'd appreciate it - it's hard to find.
Comments
All fields are optional, email address will not be shown; no HTML, URLs are automatically hyperlinked.
The Open eBook Publication Structure Working Group started tackling this problem right before the OEBF imploded it. I think there may still be a draft standard out there, though I don't know if that's where Adobe got the idea. Ping me by email if you like, and I'll check my ancient archives on my home machine.
See also http://cavlec.yarinareth.net/archives/2004/06/09/the-real-pink-number-problem/ and http://cavlec.yarinareth.net/archives/2003/02/10/hypertext-and-indexes/
while it loses the advantage of having the starting points implied by where the annotation sits, why not keep the adobe method and put it at the end of the document?
Michael: that could be good. I guess the benefit of having it inline is that it's less susceptible to breaking if the structure of the document changes: you wouldn't have to recalculate the pointers every time. Maybe that's not enough of a problem though.
One benefit of having the annotations separate on the other hand, as Adobe point out, is that you can share them with other people.