Re: [topicmapmail] XML, Topic Maps and Everything (was Re: [xml-dev] Rel

["Andrew S. Townley" <ast@atownley.org>]

Hi Richard,

On 8 Dec 2010, at 9:54 AM, Richard Light wrote:

> In message <58F37483-654F-479B-9515-7644C6582AF3@atownley.org>, Andrew S. Townley <ast@atownley.org> writes
>> 
>> Here are some of the initial thoughts I had on how to apply TMRM fundamentals to the problems being discussed on the list about where to go with XML.  I don't claim these are fully baked and there's likely holes large enough to drive a fleet of trucks through, but I did want to try and explain some of the things I see based on my experience over the last few years designing, implementing and using an information management system based on the TMRM.  Part of my thinking isn't "pure" TMRM and has also been influenced by some work by the authors of the specification in both implementing and explaining it to others[1], so that puts a particular slant on things as well.  Any errors in expressing the TMRM or Topic Maps are purely my own and should not be considered flaws of the specs themselves. ;)
> 
> Andrew,
> 
> Since you are using the fact that the TMRM expresses a directed graph model as a starting point, one issue that interests me is the extent to which the built-in TMRM features, such as subject identity and superclass-subclass relationships, add value that is relevant for generic information modelling. Value that can not, for example, be delivered by the widely-used RDF model.  (This isn't just relevant to the "XML 2.0" discussion; it also goes to the heart of why Topic Maps [may] have a contribution to make to the Linked Data paradigm.)

Naturally, I think they apply in a lot of different places, but I think the model can be applied in the particular domain of interoperable information exchange (which is one of the big use cases for XML) without trying to address the issues of more general knowledge/information representation space.

For what it's worth, though, I generally treat RDF graphs as a set of constraints applied to the TMRM foundation so that I can easily focus on the core of the information being represented and the structure it takes.  Others undoubtedly take a different view... ;)

> In the context of "XML 2.0", I don't see how your approach is going to apply to XML's (and SGML's) original job description, which is to mark up texts.

Markup in SGML or XML is a tree of nodes.  To me, what they contain isn't that important.  I have used my TMRM system to represent everything from a trivial address book to a rather lengthly formal architecture description for a complex software system.  The latter was primarily text, but the advantage was the ability to reference and link nodes as needed with a consistent model, whether these were inline references within the text or whether they were references to relationships between proxies that contained a bunch of text (among other things).

As a result, I think it's extremely relevant to the domain, but I'm not sure that XML can afford to be tied to a particular physical representation anymore.  There's too much conversion required in everyday systems, and the lack of nice integration of data referencing, identity and relationships to other formats and implementation technologies has always been an issue for XML.

As I mentioned above, I think XML's mission is to enable interoperable information exchange.  If that information happens to be mostly text, then super.  If not, then the same model will work.

Maybe what needs to come after XML 1.x isn't XML at all, but a specialized domain model for representing information for particular use cases.  However, I think that ruling out that kind of thinking wouldn't be in line with some of the other sentiments expressed and straw-man proposals I've seen so far come out of this discussion.

What if you could apply XSLT templates to JSON objects directly?  How about applying XPath expressions to navigating the object graph of a running software system?

I'm not saying that these would necessarily be everyday use cases, but in some of my work, I've come pretty close do doing both of those things--just not quite so formally.

Maybe it doesn't fit everyone's idea of "XML 2.0", but I think there's a pretty decent overlap.

> Finally, in terms of offering something new for XML 2.0 (though maybe something no-one really wants ;-) ), there is the thought that if we do move towards a directed graph approach, it becomes possible to address the "multiple views of a document" (aka overlapping hierarchies) issue which was last supported by the SGML CONCUR feature.


Naturally, and I do this regularly with my TMRM-based system.  It generally works quite well, but it depends heavily on a consistent addressing and identification model.  Fortunately, those come along as part of TMRM and the Topic Maps specifications. :)

Thanks for your comments.

ast
--
Andrew S. Townley <ast@atownley.org>
http://atownley.org