It's important to not focus overmuch on the syntax of things like Topic
Maps (or RDF or HyTime or Xlink or ...). If you think of topic maps in
terms of their fundamental abstract data model, then it becomes clear that
a relational or object database implementation must be sensical as long as
it reflects the same data model.
As part of the topic map design work, I developed an EXPRESS-G model of the
topic map data model. Once we've verified this model (it was just my own
effort done largely to help me understand the topic map design), I plan to
post it somewhere (probably the HyTime User's Group site, www.hytime.org).
Taken as input, any compentent database wonk could use that model, plus the
prose of the topic map standard, to implement a provably conforming topic
map processor. The data of that processor could always be interchanged
using Topic Map syntax.
As for implementatons, I have also developed a sample Topic Map using Xlink
and displayable with Fujitsu's HyBrick tool, which implements Xlink
sufficient to allow navigation of the links. I built a style sheet that
takes advantage of Topic Map-specific relations to present information
about the topics and relations over and above there base linkish nature.
This I will also post once I've verified that it's reasonably correct. I
also plan to implement a simple Topic Map engine on top of PHyLIS once I
complete my current reimplementation effort (hopefully by the end of the
year).
Cheers,
E.
-- <Address HyTime=bibloc> W. Eliot Kimber, Senior Consulting SGML Engineer ISOGEN International Corp. 2200 N. Lamar St., Suite 230, Dallas, TX 75202. 214.953.0004 www.isogen.com </Address>