Ethnoclassification and vernacular vocabularies

The latest meme to catch fire in the IA community deals with the folk classification tools found on systems like del.icio.us and Flickr. Users are able to freely tag content with whatever metadata comes to mind.

Headshift provides a good overview of the issues at hand, and Alex does his thing when making sure we appreciate the good and the bad of such approaches.

I just IMed with Victor about this for a while, and thought I’d chime in with my perspective.

I touched on this topic almost three years ago in a post titled “Vernacular Thesauri,” based on a talk I saw at the 2001 ASIS Annual. Go read it. It’s about porn.

Okay, you back?

First off, I think we should drop the term “folksonomy.” No offense to Thomas — it’s a catchy term, which, I guess, is why it has caught on. It’s also inaccurate. What bugs me most is the use of the word “taxonomy.” Taxonomies tend toward hierarchy, and they tend to be imposed. Tagging does not a taxonomy make.

What we’re talking about here is “classification.” In rooting around, trying to find some prior research on this topic, I plugged “folk classification” into Google, it turns out that anthropologists have done some thinking around this, particularly with respect to ethnobiology, or how the folk approach biology, and ethnoscience.

This lead me to think that the appropriate term would be “ethnoclassification”, and when I plugged that into Google, I found “Slouching Toward Infrastructure”, a page for a 1996 Digital Libraries Workshop lead by Susan Leigh Star.

The practice of tagging on del.icio.us works because, at its heart, it’s meant for the use of the individual doing the tagging. The fact that it contributes to the group is a happy by-product… But as a tool for group tagging, it’s woefully insufficient. Del.icio.us has a very low findability quotient. It’s great for serendipity and browsing, and an utter disaster for anything targeted.

This is where Alex’s quest for a middle ground resonates with me. Being as wedded to the practical as I am, I wonder how we could put such ethnoclassifications to work in useful contexts. I’m thinking maybe an intranet, where people are free to tag documents as they see fit, but there is some librarian/IA role that attempts to provide some degree of robustness to such a scattered classification. If nothing else, this approach would be a boon to developing thesauri, particularly variant terms.

4 thoughts on “Ethnoclassification and vernacular vocabularies

  1. I’ve been vaguely thinking around the idea of doing some research on an intranet project, looking at how people classify their email and the labels that they use. I’m certain that there would be a bunch of interesting consistency and a lot of interesting differences. I wouldn’t use it as a research – solution step, just another way of looking at how a boundaried group describe a consistent bunch of material…

  2. Sebastian Paquet’s *Using an Ontology to Communicate Knowledge Across Disciplines* (http://radio.weblogs.com/0110772/stories/2002/08/05/usingAnOntologyToCommunicateKnowledgeAcrossDisciplines.html) may also be a decent resource. I have only browsed through the article and will be reading more fully on the train today.

  3. Don’t sweat it. I’m really only getting started on del.icio.us and will be adding a lot more to enhance findability and discovery.

    I’ll be adding a bunch of magic to tie all the different bundles of related terms together. Each user should use whatever they feel they are most comfortable with and I will do the rest behind the scenes.

  4. …seems to me like “folksonomies” are just one way of widening the size of the committee that’s deciding on the ontology to use… and hopefully using some automata to moderate the community.

    The reasons why “ontologies” are so annoyingly hierarchical and fixed is that they are *simplifications* of language, destined to be used by machines (at least, in the latest incarnation of the term). So *someone* has got to put limits on the vocabulary. Instead of a small group of people, here you get a community to vote by consensus. You still end up generating an ontology at the end of the day.

    The real issue, it seems, may be how to make ontologies adapt with time- for example, by developing evolving metathesauri?