W3C home > Mailing lists > Public > public-swd-wg@w3.org > February 2007

Re: [SKOS] Re: more SKOS requirements

From: Daniel Rubin <rubin@med.stanford.edu>
Date: Mon, 12 Feb 2007 10:37:13 -0800
Message-Id: <>
To: Antoine Isaac <aisaac@few.vu.nl>
Cc: SWD WG <public-swd-wg@w3.org>

Hi Antoine,
see below:

At 07:55 AM 2/12/2007, Antoine Isaac wrote:
>Hi Daniel,
>Thank you for the updates! Before trying to answer you question 
>about the link between OBI and Radlex, some content-related questions:
>Reading Radlex description, I wondered how the link between a 
>special relationhip like "part" and "parent"/"children" is dealt 
>with. Something like
>- a SPARQL query (when I ask for children, give me the parts)
>- RDFS subtyping (part rdfs:subPropertyOf child)
>- or dynamic interface generation (the "parent information not being 
>encoded as such in the model, but derived when asking a 
>visualization of the concept)?

If you are asking about how queries will work on RadLex, it will be 
via either SPARQL or a custom interface to an application that 
consumes the terminology.

>Also I wonder whether a special relationhip can always be converted 
>in parent/child link. There are some links, like "Continuous with" 
>that do not seem to be interpretable naturally as hierarchical ones. 
>But I am not sure, and could not find any example in my 5 minutes browsing...

This is basically the issue that RadLex is an ontology and not a 
simple terminology. "Hierarchy" depends on the relationship used to 
construct it. So RadLex actually has multiple hierarchies.

>Finally, what is the kind of mapping you envision/are developping? 
>"Just" equivalence or subclass? Or more complex links?

Two initially-- synonymy and subclass. Later, we expect people to 
compose atomic terms (post-coordination) to describe composite entities.

>PS: about the OBI description, wasn't there a small typo problem? 
>What I have in the "Standards and guidelines considered" include:
>"We is the use ofan an alphanumeric identifier for the term versus 
>using the human readable term name. The reason for this was to 
>remove semantics from the unique identifier for the term.render the 
>identifier unique and ease term obsoletion and versioning."
>>Hi Antoine,
>>I added the OBI metadata requirements to the corresponding case description.
>>I also posted my edited UCR for Radlex.
>>At 10:30 AM 2/10/2007, Antoine Isaac wrote:
>>>Hi Daniel,
>>>>A few of the communities I've been working with for SKOS use 
>>>>cases have come together to identify a set of metadata elements 
>>>>they need to be associated with terms in their published 
>>>>terminologies. I feel these should be considered for additional 
>>>>SKOS requirements.
>>>The metadata elements you give are indeed now considered (at least 
>>>the general categories they belong to) in the candidate 
>>>requirement list, or in the issue list.
>>>But of course the items there need further definition. Alan has 
>>>accepted to do some investigation on this:
>>>>ACTION: Alan to write down the general documentation 
>>>>requirements, in particular to those that are related to literal 
>>>>values, and how to represent that in skos [recorded in 
>>>>ACTION: Alan write up the preferredLabel modelling issue 
>>>>[recorded in http://www.w3.org/2007/01/22-swd-minutes.html#action01]
>>>I think definitively you should collaborate together on that! Your 
>>>OBI list is a good input. By the way, I think it should appear 
>>>explicitly in the corresponding case description, shouldn't it?
>>>Thanks for your input,
>>>>Here is a list of metadata properties and definitions:
>>>>This list was compiled over the last months, starting from the
>>>>RU_metadata.owl, the BirnLex metadata requirements and the metadata
>>>>annotations provided by the NCIT (all of which can be found on the
>>>>OBI-Wiki at
>>>>This minimal subset of metadata properties was agreed upon, and had its
>>>>definitions finalized by the OBI metadata subgroup during the last two
>>>>The added cardinalities are still under discussion, as is the
>>>>implementation and in some cases the domain of the metadata properties.
Received on Monday, 12 February 2007 18:37:22 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:07:49 UTC