Re: Data Display & Modeling

"Dawn M. Wolthuis" <dwolt_at_tincat-group.com> wrote in message news:c7enir$d77$1_at_news.netins.net...
> I did a very quick search on google for
> data-model di-graph
>
> and found several pieces, such as
> http://www.dcs.bbk.ac.uk/~mark/download/kde.pdf
>
> Are folks like these as clueless as I am? --dawn

The first sentence of the abstract: "Currently database researchers are investigating new data models in order to remedy the deficiences of the flat relational model when applied to non-business applications."

First of all, "flat" is somewhat argumentative in an abstract, since relational is only as flat as any 2-D projection of an N-dimensional space.

Second of all, "non-business applications"? That's all the new data models are good for? What exactly does "non-business" (or business, if you like) mean anyway? The examples they give seem to indicate large data structures best defined as types of their own, or perhaps decomposed into useful types (I'd be surprised if a chip design couldn't be!) and relations (predicates about values of those types). The issue of text searching is a little more complicated, because they're less specific, but I'll certainly admit that XPath and XQuery are useful operators over an XML type - NOT an "XML database", whatever that might be.

I may read the paper, though my stack is really large as I lack the restraint to not read most of what I stumble across in Citeseer.

I'm not claiming these researchers are clueless (nor that you are), but it's a weak opening, a quick pass over some weary arguments before they dive into their topic. Formalisms for XML and graph-based approaches are popular, but in research as in all things, that doesn't mean useful. "Important" is debatable.

As always, I appreciate your raising these points. I do think it's useful.

• Eric