Flickr releases interestingness and clustering
Oh dear. As if I need any new Flickr goodness to distract me from catching up after returning from BlogHer. Flickr has gone and released interestingness, which uses an algorithm full of "a whole bunch of secret sauce" to analyze user behavior around photos and present the most interesting in a given time period. For example, check out the most interesting photos from the past 24 hours.
As if that weren't enough, they've also introduced clustering, which is a new way to explore some of the finer distinctions within tag clusters -- e.g. the nature cluster is broken down into sets such as:
- water, lake, river, reflection, rain, bridge, view, boat, drops, window
- tree, trees, sky, sunset, blue, clouds, leaves, leaf, forest, sun
- flower, flowers, green, macro, landscape, yellow, red, garden, spring, plant
[Via danah boyd]












Comments
2
Subscribe to commentsmail@coreyh.comAug 2nd 2005 1:19PM
I am hereby wishing and hoping for an RSS feed for interestingness. I refuse to bookmark.
Barb DybwadOct 7th 2005 4:41PM
I'm not sure an RSS feed would work so well, actually. It's just that "interestingness" isn't tied to a particular timeline -- a photo's interestingness changes over time as users interact with it. You could maybe track a timestamp of when each photo first becomes "interesting," and get an RSS feed from that... but it's sort of missing the point that user activity over time affects relative interestingness. RSS would have a really hard time capturing that.