Lately I've been building a photo collection that's referenced to "linked data" terms, which, for the most part, correspond to pages in wikipedia.
I've developed some other sites based on a process that uses flickr and paid crowdsourcing, but I was having trouble with people who thought that the NYSE was the "American Stock Exchange" -- so I got the idea of building a database of reference images from wikipedia + commons... That turned into a bigger project than I expected, and I figured that I might as well (a) put those images in my collection, and (b) develop my import system to the fullest extent.
Right now,
is a photo-geo mashup that combines photos from Wikimedia Commons with maps from Open Street Maps. Although it's not hard to find coordinates that are off, it provides a really fun browsing interface.
We're working on improving the browsing interface, doing data cleaning and such -- for at least the six months, I think we'll still primarily getting images from Commons, but we eventually expect to restart an improved version of our old pipeline, as well as allowing volunteers to contribute content. I hope that there's a day that Commons will be getting images from Ookaboo rather than the other way around.
Any thoughts?