[Sigia-l] tagging software

Ziya Oz listera at earthlink.net
Mon Mar 3 13:24:59 EST 2008


Frank Shepard:

> if the context is a single user and her data, why not allow a
> machine to generate tags? What's the objection to that?

When humans tag they make a cognitive judgment on a doc in a way we don't
always fully understand. When machines do the tagging, we tell them how to.
Machines can arrive at a tag in many ways: it could be simple concordance
parsing or complex semantic relationship discovery. We can also allow
machines to go out and interface with other datasets, knowledge repositories
and compare other tagging structures, etc. Humans can theoretically do some
of  these, but they usually don't, certainly not at a scale and volume
machines can manage.

I'm not aware of a controlled experiment where machines and humans 'tagged'
the same body of docs over a meaningful time span to see
differences/similarities. I recon there'll be some.

Ziya
Nullius in Verba 






More information about the Sigia-l mailing list