[Sigcr-l] question on The Power To Tag 12|03

[Hope A Olson]

I entirely agree. You've hit on one of my favorite soapbox topics.

Stated somewhat simplistically, a taxonomy is a structure made of paradigmatic relationships between concepts - those concepts that are in some essential manner always linked such as "dogs" and "poodles." Authority records document paradigmatic relationships. However, a "folksonomy" is really a set of syntagmatic relationships between those concepts that occur together in a particular context but are not essentially linked such as dogs and grooming. Some dogs get groomed, others don't; some acts of grooming are done to dogs, others are done to children or lawns or oneself. There are also inbetween relationships in which the syntagmatic becomes so common it verges on paradigmatic - like "dogs" and "walking." Free text is generally full of syntagmatic relationships. The co-occuring thesaurus descriptors assigned to a database entry are syntagmatic. Precoordination introduces syntagmatic relationships to controlled vocabularies. LCSH subdivisions generally have a syntagmatic relationship to the main heading - "Dogs--Periodicals" - there is no essential relationship between the two concepts.

A folksonomy is not entirely unstructured. Tagging by multiple taggers ends up with not only individual tags, but also clusters of tags that occur in a power law type of distribution. That is for any one item tagged by multiple taggers there will be a core of frequently used terms diminishing to a long tail. The same will be true of pairs of terms. However, the relationships even in these pairs are still syntagmatic - there seems to be no taxonomy in folksonomies.

However, I would argue that there are paradigmatic relationships in folk taxonomies. Within a given culture, the relationships are essential - a certain kind of bird is related to a certain group of people as is a certain compass point and a certain season. These are all part of a structure that, though culturally developed, is paradigmatic within its own context. If one conceives of social tagging sites as potential cultures then it is possible that a folk taxonomy could evolve (maybe it already has within some groups - it would not surprise me and I'd be interested if anyone has examples), but it will have to go beyond simple co-occurrence to establish relationships at the level of conceptual essence.

One could go on to link paradigmatic relationships to warrant and bring in classifications of social and natural phenomena compared with bibliographic classifications, but I'll stop there.

Hope

Hope A. Olson, Professor and Associate Dean
School of Information Studies
510G Bolton Hall
University of Wisconsin-Milwaukee
Milwaukee, WI 53201
USA
http://www.uwm.edu/Dept/SOIS
email holson at uwm.edu

----- "Simon Spero" <sesuncedu at gmail.com> wrote:

> From: "Simon Spero" <sesuncedu at gmail.com>
> To: "Margaret Kipp" <margaret.kipp at gmail.com>
> Cc: sigcr-l at asis.org
> Sent: Friday, November 27, 2009 11:45:31 PM GMT -06:00 Guadalajara / Mexico City / Monterrey
> Subject: Re: [Sigcr-l] question on The Power To Tag 12|03
>
> On Fri, Nov 27, 2009 at 9:50 PM, Margaret Kipp
> <margaret.kipp at gmail.com>wrote:
> 
> > There has been one review article of tagging research published so
> far that
> > covers research up to 2007.
> >
> > Trant, J. 2009. Studying social tagging and folksonomy: A review
> and
> > framework. Journal of Digital Information 10(1).
> > http://journals.tdl.org/jodi/article/view/269
> >
> 
> I still think that folksonomy distracts us from the importance of 
> "folk
> taxonomies" (as contrasted with  "scientific taxonomies"). Using
> Hayek's
> terms, Folk taxonomies are  *Kosmos* , or grown order; scientific
> taxonomies
> are *Taxis*, or made order. By virtue of being culturally evolved,
> Folk
> taxonomies tend to be a very good match with human cognitive
> constraints,
> including  the primacy of  the basic level terms and a tendency to be
> much
> shallower than artificially designed orderings (I seem to remember
> reading
> that they typically had a  maximum depth of  7±2 , but I forgot to
> mark the
> spot. Stupid non-greppable dead trees).
> 
> I have 157 tagging related articles tagged on CiteULike if anyone is
> > interested (http://www.citeulike.org/user/meikipp/tag/tagging)
> > including about 125 that I cited in my thesis.
> >
> 
> If one  subscribed to the RSS feed, one could even have inferred that
> you
> were about to defend based on the rate that entries were being added
> this
> summer :--)
> 
> Simon
> _______________________________________________
> Sigcr-l mailing list
> Sigcr-l at asis.org
> http://mail.asis.org/mailman/listinfo/sigcr-l