The Semantics of Tagging - Social Semantics: The Search for Meaning on the Web - page 107

Yet, what are the semantics of a tagging system? A number of problems stem
from organizing information through tagging systems, including ambiguity in the
meaning of tags and the use of synonyms which creates informational redundancy.
Interestingly, Semantic Web ontologies like NiceTag have been developed to address
the issues of ambiguity in tagging systems by formalizing the tagging process itself,
often by linking a particular tag to a Semantic Web URI (Monnin et al. 2010).
While this may clarify the intended meaning of the tag, this approach does not
thereby in some semi-magical manner give semantics to the tag. Also, it seems the
most interesting question for our approach is not what the referent of a particular
tag act, but whether or not the collective sum of individual tagging acts can serve
as an objective notion of sense. Since each tag for a given web resource (such
as a web-page) is repeated a number of times by different users, for any given
tagged resource there is a distribution of tags and their associated frequencies. The
collection of all tags and their frequencies ordered by rank frequency for a given
resource is the tag distribution of that resource, which is our candidate for a Fregean
sense.
So then, the important open question concerning the use of collaborative tagging
to organize metadata is whether the system becomes stable over time. By stable ,we
mean that users have collectively developed some implicit consensus about which
tags best describe a site, and these tags do not vary much over time. Only this will
allow tags to be used as an adequate computational theory of neo-Fregean sense,
since otherwise tagging would be subjective rather than objective. We will assume
that these tags that best describe a resource will be those that are used most often,
and new users mostly reinforce already-present tags with similar frequencies. Since
users of a tagging system are not acting under a centralized controlling vocabulary,
one might imagine that no coherent categorization schemes would emerge at all
from collaborative tagging. In this case, tagging systems, especially those with
an open-ended number of non-expert users like del.icio.us, 1 would be inherently
unstable such that the tags used and their frequency of use would be in a constant
state of flux. If this were the case, identifying coherent, stable structures of collective
sense produced by users with respect to a site would be difficult or impossible.
The hope amongst proponents of collaborative tagging systems is that stable tag
distributions, and thus, possibly, stable categorization schemes, might arise from
these systems. Again, by stable we do not mean that users stop tagging the resource,
but instead that users collectively settle on a group of tags that describe the resource
well and new users mostly reinforce already-present tags with the same frequency
as they are represented in the existing distribution. Online tagging systems have
a variety of features that are often associated with complex systems such as a
large number of users and a lack of central coordination. These types of systems
are known to produce a distribution known as a power-law over time. A crucial
feature of some power laws - and one that we also exploit in this work - is that
they can be produced by scale-free networks. So regardless of how large the system
1 http://del.icio.us , which as of 2011 redirects to http://delicious.com/