geni schrieb:
On 3 February 2010 21:54, Daniel Kinzler daniel@brightbyte.de wrote:
geni schrieb:
What are you looking for that hitting [[special:random/file]] as many times as images are needed won't provide?
That would give a fair sample, but not necessarily a good sample. the usability folks need "a few of each kind" for testing. getting numbers proportional to the actual "population" of images is unnecessary and probably rather counterproductive.
So, I see two tasks:
- identifying relevant "group" of media (along the parameters guillom specified)
- sampling from each group
-- daniel
At a guess I'd say the museum categories are the best bet. Stuff like:
http://commons.wikimedia.org/wiki/Category:Science_Museum_%28London%29 http://commons.wikimedia.org/wiki/Category:British_Museum
I think the museum categories are *one* interesting group of pretty uniform content.
-- daniel