A lot of great tools exist to annotate collections and bibliographies – Zotero being one of the best lightweight examples for end-users.  At the same time, some large scale projects are exploring annotations as low-level data objects.  I want to discuss the middle -  potential annotation frameworks that could slip easily into the services layer of web applications for manipulating textual collections, particularly TEI.  One idea is to use AtomPub to post, retrieve,  and edit annotations tied to texts and text collections. There are several benefits to this approach:  one is the ease with which one could embed metadata that could be used to ingest annotations into a digital repository as independent objects, to be recombined with texts at the application level;  Another is that it would establish an annotation framework that could apply to diverse types of collections, and would enable the ability to annotate data using rich media.

While AtomPub is easy to implement, building connections between Atom documents and very granular segments of text or multimedia is more difficult.  For TEI, there are some native tools (XPointer), but they are fairly clunky.  There are also abstraction tools that could be used to tokenize a text for annotation purposes, but the complexity involved in building that abstraction layer may negate the benefits of a simple,  RESTful annotation framework that uses AtomPub.

I would like to work with other folks at THATCamp to brainstorm and hopefully test some ideas for using AtomPub for granular annotation.