W3C home > Mailing lists > Public > semantic-web@w3.org > March 2005

Re: "Tag" metaphor and "Tag User Interface"

From: Richard Newman <r.newman@reading.ac.uk>
Date: Thu, 24 Mar 2005 11:07:18 +0000
Message-Id: <63716cc205b00b8de97e1a5452457832@reading.ac.uk>
Cc: semantic-web@w3.org
To: Max Voelkel <max@xam.de>

Max,
   That's an interesting approach; I (and I think Morten's WordPress 
plugin and Annotea?) was going the opposite route, minting a URI for 
each tag, but going this way is intriguing (if you have the existing 
vocabularies).

   I do see a problem or two.

WORKS:    "somepaper" "creator" "Richard Newman"
           =>  <.../somepaper> dc:creator "Richard Newman"
           I expect you fall back to literals?

           "somepaper" "maker" "RichardNewman"
           =>  <.../somepaper> foaf:maker rich:RichardNewman
           ... assuming you know my FOAF file.

DOESN'T:  "somepaper" "author" "Richard"
           ... doesn't map to any existing URIs.

   Also, it's not quite tagging as such... I think you're right to call 
it a "tag user interface". One of the values of tags is the ability to 
clump together a whole lot of ambiguous meaning into one word --- such 
as my "Highlander" tag. It's very much a personal statement that only 
has real meaning in my context, and doesn't expand out into just one 
statement.

   I could see some kind of templating mechanism that works like this:

1. User enters "Highlander"
2. "Highlander" matches <.../pubs/Highlander> and 
<.../films/Highlander>.
3. User picks one; the pubs one has an "(photo of) event that took 
place at" template.
4. RDF is expanded into a subgraph.

   That's pretty much what I try to do in my iPhoto plugin, but I let 
the user do the mapping once, from templates, then save the mappings 
between sessions (so "Highlander" is remembered as a place annotation).

   -R

On Mar 23, 2005, at 19:59, Max Voelkel wrote:

>
> Hi all,
>
>   most ontologies I have seen use some namespace, then a hash or slash
>   and then more or less englisch terms. These last parts look like
>   tags, the first part - the namespace - is then the context.
>
>   I am right now designing a wiki system that lets the user enter RDF
>   statements in a notation similar to
>     tag   tag   "a longer tag"
>      (s)   (p)       (o)
>
>   I use
>   "any text" "other text" "more text"
>   and upon page save in the wiki each string is given a unique uri.
>   This creates rdf like this:
>   _:1 rdf:value "any text"
>   _:1 rdf:value "other text"
>   _:1 rdf:value "more text"
>   _:1 _:2 _:3
>
>   If a string is the value of two or more uris, the user is prompted
>   to choose among the available uris. I think this could lead to a
>   very easy to use user interface without loosing "uri-precision" in
>   disambiguating terms. Probably such string-to-uri service should  be 
> a REST-ful
>   webservice...
>
>   Ok, that's I all, I just wanted to make the point that tagging not
>   neceassaryly means ambiguos data.
>
> Regards,
>
>   Max
> --
> University of Karlsruhe, AIFB, Knowledge Management Group
> room #258, building 11.40
> mvo@aifb.uni-karlsruhe.de   +49 721 608-4754   www.xam.de
>
>
>
Received on Thursday, 24 March 2005 11:07:55 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:45 UTC