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

Re: tagtriples

From: Phil Dawes <pdawes@users.sf.net>
Date: Wed, 23 Mar 2005 20:13:10 +0000
Message-ID: <16961.52694.190757.25324@gargle.gargle.HOWL>
To: Jeremy Wong <50263336@student.cityu.edu.hk>
Cc: semantic-web@w3.org

Hi Jeremy

Jeremy Wong writes:
 > I like the user interface. It is simple to use. Any entailment does the 
 > tagtriples support?

Not yet - I'm not sure how comfortable I'd be with entailing things
from tags, given the potentual ambiguity of tags. I suppose it would
have to be tightly controlled by the person doing the aggregating.

Having said that, the query code is pretty similar to my veudas rdf
store code (which supports forward and back-chained reasoning), so
adding this functionality would be straight-forward if it turns out to
be required.

To be honest, things we used entailments for at work are mostly not
required in this system. We tended to use

- fc entailing subproperties of rdfs:label 
(not required, since tag can be displayed as the label)

- joining data using owl:sameas and owl:inverseFunctionalProperty
This was used mainly because people used different URI schemes to
represent the same things. With the tag stuff the data tends to stick
together more easily because the data agrees on well-defined things.
(e.g. machine names, ip addresses, application names, people)

If anything, we are likely going to have the opposite problem -
pulling apart data that shares tags that don't mean the same thing.


Received on Wednesday, 23 March 2005 20:12:59 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:47:01 UTC