[Sigia-l] Research papers on tagging

[Ziya Oz]

Stephen Collins:

> His argument is that given  everything is searchable, everything automatically
> becomes a tag he  can use and tagging is therefore, noise.

It all depends on how search is defined/conducted. There's a huge
difference, for example, among: simple concordance search, self-learning
Bayesian search, PageRank'ed search, social-network'ed search, custom search
with entity extraction, etc.

It's entirely possible to devise a search system that does unambiguously
better than a human-tagged system for some volume of data/info/docs.

> - user-defined meaning adding value to formal meaning

This could be bad if the tagger/user is not competent; algorithmic system
can do the same or better.

> - emergent context and meaning

Ditto.

> - the spurious nature of the every word becomes a tag position based
> on my position that to do so simply creates confusion rather than
> meaning

"Search" doesn't  have to be this pedestrian.
 
So this argument really depends on the what the meaning of 'search' is,
mostly because semantic approaches and algorithms for contextualizing and
inter-relating have progressed quite a bit. It's amazing what those computer
thingies can do, given a half-smart human to design them.

Ziya
Nullius in Verba