[Sigia-l] tagging software

[Ruth Kaufman]

On Mon, Mar 3, 2008 at 1:07 PM, W Evans <wkevans4 at gmail.com> wrote:

> There are certainly a couple (mostly in very very early stage
> development),
> that will do "tagging" of a person's content by the machine once it's in a
> machine readable format. But it's not called tagging, it's called semantic
> analysis -


In your view, would a content owner or someone in an editorial role applying
a tag, as opposed to an end user, be considered tagging? Does it make a
difference if the tag is applied in the CMS or in the end-user experience?
Does it matter if the values of the tags are selected from a controlled
vocabulary versus free-form text?

I always considered tagging to be the act of applying tags, regardless of
the technology or process used to do it. I think what you're trying to get
at is the underlying reasons people (users or content / application
managers) might use tags -- for general classification, for content
management functions, to trigger a memory, to serve as a sort of bookmark,
etc. I don't think any one of these concepts can own or dominate the notion
of tagging content.

I think what distinguishes tagging from, say, semantic analysis, is that
tagging involves a data element called a tag -- usually in the form of HTML
metadata or a specific XML element that is, for all intents and purposes,
used as metadata. In contrast, a machine can do semantic analysis but never
actually apply tags -- it can store the analysis in an external resource,
for example. If the machine does append a tag to the content resource, then
it has simply selected the tag value using an algorithm, but ultimately, it
has tagged the content.

Why does this matter? Tags, once applied, are descriptive, static content
and often made available for applications to reference, whether behind a
firewall, through APIs, or in a social media application. A findability and
classification alternative is not to depend on tags but to analyze the
content (with machines or humans) in real time and decide how to treat it
based on the results of the analysis, rather than based on a pre-defined
descriptor -- for example, semantic search applications or our raw human
ability to comprehend and judge what we read and how we respond to it --
enjoy it, store it, pass it to friend, etc.

My 2 cents.

-Ruth