[TagCommons-WG] Use Cases for Sharing Tag Data - Summary

[Tom Gruber]

Hello,

We had a great response to the call for use cases -- we also were introduced
to some of the existing applications and other activities in this space.
Here's a summary of the use case discussions, which I will also put onto the
tagcommons.org blog after a quick review period on this list.

Use Cases for Sharing Tag Data

1. Personal bookmarking across tagging sites (where all objects are web
links)

2. Browsing and searching using other people's tag data for research and
discovery - across sites and sources (like pivot browsing del.icio.us to
find related sources on a topic, but across multiple tag sources)

3. Improving signal-to-noise in search by aggregating tag data from multiple
people across sites (eg, collaborating filtering to beat spam)

4. A reference service to cross index objects across disparate applications
with differing data types, sources, and media (eg, books, pictures, web
bookmarks, people related to a topic)

5. Organizing documents on personal file systems or shared repositories
using tags (where the file systems or repositories are different
applications, such as document management systems)

6. Searching and monitoring across multiple tagging services (example:
Thomas Vander Wal's Tagwatcher)

7. Distributing user-contributed content (eg, reviews) with tags on the
Semantic Web (Tom Heath's revyu.com)

8. Comparing tagging behavior across applications, communities, languages,
scales, etc.

Functional requirements that came from this discussion or related reading:

- a way to get tag data from multiple sources, indexed by tagging person,
tagged object and tag name

- a way to match/intersect tag assertions on tagging person, tagged object,
and tag name

- information about the source or venue or community scope of tag, to help
disambiguate matches

- the date of a tagging assertion

- the natural language of the tag word(s) in the tag assertion

- matching personal identity across applications / venues

- matching tag words across variants of spelling, synonymy, etc.

- exposing tag bundle data (using other tags on the same object by the same
tagger as disambiguating context)

What did I leave out?   Are these coherent?  Please comment and I'll write
up a clean document for the blog.

Tom