Fantastic article by Golder & Huberman from 2005 observing patterns of tag use on del.icio.us. All of it is good, including a great discussion of various forms of ambiguity etc in tagging and the relative merits/demerits of taxonomies vs folksonomies, but the really interesting bit comes towards the end when they observe, and attempt to account for, a remarkable and stable consensus on proportions of tags used to categorise resources. They relate this to the stochastic urn model, observing that some of the stability is the result of imitation (what I have called mob behaviour) but some of it is to do with shared and presumably independently held knowledge (what I have called multitude behaviour).
Created:Thu, 31 May 2007 05:22:07 GMT

Original: http://jondron.net/cofindfrshowresource.php?tid=5325&resid=1289
Posted: May 30, 2007, 11:22 pm