TEI Stand-Off Markup WG 2002-10-04 Conference Call Notes

NI draws difference between SOM pointing at elements or segments that are designed to be easy to point at, and using ranges or regions so that markup is not needed in the ‘primary’ document.

NI claims there is a limit to how well you can mix the two approaches.

NI's group wants the information associated with a ‘segment’ (a range of text marked up in the primary document so that it is easy to point out by SOM) to be pretty minimal, with most semantic information in the SOM.

NI [?...?]

DD points out that TEI texts may well be created for purposes other than linguistic analysis, then used for linguistic ana. Thus proposes we suggest ‘linguistic ana in separate layer’ rather than ‘don't use inline [editorial] elements’.

CL points out ATLAS uses all SOM. Source primary data is treated as read-only.

JH describes advantages of in-line over stand-off for a R/W text that changes.

CL describes ATLAS's system.

NI asserts that the in-line and stand-off methods really do the same thing: say ‘here is the beginning’ and ‘here is the end’.

DD asks of ATLAS DTD — looks like there's not a lot of provisions for hierarchical relationships between parts of the annotations.

JC describes Knight project which deals with discontiguous elements; time is inherited in some cases and not in others. Trying to build data model more general than annotation graphs. Element can point to another in a named role, but may or may not inherit time from children.

CL: ATLAS can support an assigned role to each annotation. Because newly created objects need to be validated in memory anyway, have not put any effort into developing a schema for XML validation. ATLAS has developed its own system for ontologies, does not use RDF.

DD would like us to continue this discussion on the list; SB requests people even post summaries of what they've said today.