Craig Franklin wrote:

But to show how this is inadequate, can you produce me a list of... say... Polish-born Jewish authors who died in the United Kingdom in 1897? That's what I feel this tool would be really useful for, and the existing categories system is pretty much inadequate for that (I suppose if you looked at all of the relevant categories and returned a list of links that appeared in EVERY one that'd work, but there's currently no tool for this, and I imagine that it'd be fairly expensive in terms of CPU cycles.

I suppose that adding a tool that would do what I proposed above would be an acceptable compromise, but I figure if we're going to do such a thing, then we may as well do it properly ;-)

Yes. Our search system was written when the 'pedia was much smaller (less than 10% of its present size). I suppose that our developpers must be aware of the needs of our more sophisticated information pile. They just need a bit of free time to do it. :-)

Oh, I didn't mean it in the sense of "we should invent our own metadata

tagging scheme". Such a thing would be overkill. I think the best solution would be to have a relatively simple tagging scheme like this:

[{Died: 1945}] [{Nationality: Dutch}] [{Religion: Jew}] [{Profession: Author}]

...and then automagically convert that into RDF, or some other mutually agreed-upon metadata scheme. That way casual writers don't have to worry about the intricate details of such a system, and anyone who wants to get at the actual real RDF can, presumably through a direct database call of some description. Of course, I know very little about RDF or how it works, so I don't know how much of a problem automated translation like this would be.

[[Category: ]] should be quite adequate for these. Introducing an open-ended system of name-spaces gives that many more opportunities for things to go wrong. [[Category:d1945]] would be better than [[Died:1945]]

Your post also brings up another interesting point - the metadata will have to be almost or totally complete to be useful. There's little point in doing searches on "Author" if only 20% of the authors in the database are tagged as such. How we can quickly go through a half-million articles quickly and tag them is beyond me.

Everything takes time. Since categories were introduced some contributors have shown great enthusiasm for the simple task of adding categories. It's a fast and easy way to build up one's article count. If your idea were implemented this would only be a temporary problem.

Ec