On 18/12/05, Andrew Gray shimgray@gmail.com wrote:
We can catch another batch by searching anything with a bio-stub tag and categorising them, then rechecking. Anything else we have no real way of finding, since there's nothing to define an article's content without actually reading it.
It's a bit silly isn't it, that the stub sorting project is doing better at categorising stubs than the rest of the project is at categorising articles? I'm not sure exactly how many stubs vs articles there are totally, but that sort of metholodogy, where they try to generally categorise what sort of topic something is in, for everything, rather than randomly adding categories like "Category:People who met JFK" and suchforth, seems to be working really well.
So I wonder if we could find a way to use this sort of methodology, of sorting things into more specific categories when needed, for regular articles. Obviously this would be done as well as the existing bottom-up approach.
But it would be a really useful feature to be able to tell progamatically whether an article (any article) is about a place, or a person or a group or a concept, or whatever. And the current model has no way of doing that - even recursively looking inside [[Category:People]] doesn't work because then you end up finding [[Category:John Lennon]] and then get to things like [[251 Menlove Avenue]], which is evidently not a person.
-- Abi