Machine Tags: Theory, Working Code and Gotchas (and Robots!)
Aaron Cope, Stamen Design, USA
Machine tags are just like regular tags with a special syntax to denote a faceted relationship: a namespace (or a subject domain); a predicate (or a subject topic); and a value.
Machine tags provide just enough structure to define and expose faceted metadata without all of the friction that prevents traditional approaches from achieving widespread adoption. Although they were first formally recognized, indexed and made searchable by Flickr in 2007 they were already being used by its userbase in an ad-hoc manner to build on and extend the service in new and interesting ways.
They are the sweet spot between formal taxonomy and the perceived mayhem of uncontrolled "folksonomies". They are not the Semantic Web, although they could play it on TV. They are small pieces (of data) loosely joined (by the Internets).
Since being formally added a first-class datatype in Flickr machine tags have been used by and for a variety of projects including OpenStreetMap, Dopplr, the Brooklyn Museum, the Encyclopedia of Life and the Royal Observatory at Greenwich.
During the workshop we will:
- Take a walk through the history and theory and benefits of machine tags and their relationship to both "traditional" tagging and formal ontologies.
- Learn how Flickr has used machine tags: search; machine tag hierarchies; wildcard machine tag URLs; and machine tags "extras"
- Cover examples of past and current machine tag integrations by both Flickr and others.
- Discuss working (and open source!) code, and a discussion of the various pitfalls working with machine tags.
- Talk about robots! (No, really.)
- Finish with a little bit of hand-wavey design-fiction around future directions and possibilities for machine tags.
Who should attend?
Anyone interested in metadata and classifications systems, anyone interested in community-driven tagging and anyone interested in the brass-tacks of how machine tags work.
In short: Anyone with an interest in the who and the how of naming things!
Machine tags are just one approach to tackling an ages-old problem but is young and has plenty of room left to grow. Come with your questions and your challenges and see what's worked so far and where it can go in the future.
Workshop Slides: http://www.aaronland.info/talks/mw10_machinetags/
Reading List [PDF]: http://www.aaronland.info/talks/mw10_machinetags/mw10_machinetags_readinglist.pdf
Keywords: tagging, machine tags