Is the bibliography trying to have a tree, e.g.: page -> article ->
journal
If it is, then *gack*. The potential for confusion is pretty large -- is Richard's metadata about the journal article or a page in the article, and how would I know?
I tend to agree with Rod on this -- but here's the thing: we want to know whether the citation is a Chapter, or a book, or an article, or a page anyway -- regardless of whether the citation is structured as a tree or as concatenated text. My feeling is: why not have both? (i.e., as Roger's vocab already has both, and whatever comes from the St. Louis meeting will, I believe, have both). I wouldn't go more than one level up: give the containing reference citation as a string (e.g., for book chapters, give the full book citation as a string), and also provide a GUID link to that containing/parent reference citation -- but don't give the full tree if there is more than one level of "parent". That is, if there is a grandparent, don't represent it via a GUID in the grandchild -- only represent the parent as a GUID, and then the metadata for the parent can itself include a link to the Grandparent.
I do believe we want to maintain the ability to walk up (and down) a hierarchy (for a bunch of reasons). But at the same time, I agree with Rod that we shouldn't expect the client to *have* to do so.
I tried to investigate this further but the Wiki
http://wiki.tdwg.org/twiki/bin/view/TAG/OntDiscussPublicationCitationPublica tionCitation
is down.
Most of the commentary ended up here:
http://wiki.tdwg.org/twiki/bin/view/TAG/PublicationCitationLsidVoc
...but even this is moot,a s it was written before the St. Louis meeting. I suggest we wait until the product of that meeting gets distributed, and then attack that strawman, instead of Roger's strawman on the LSIDVocs site.
Rich