W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > April 2005

Re: [BioPAX-discuss] LSID Best practices...

From: Sean Martin <sjmm@us.ibm.com>
Date: Thu, 7 Apr 2005 09:33:40 -0400
To: <public-semweb-lifesci@w3.org>
Message-ID: <OF9E91C3A5.7BF50328-ON85256FDC.0045D026-85256FDC.004A7F4F@us.ibm.com>
>could/should be used as RDF IDs for individuals in OWL documents?

My understanding is that there is only really one serious should with 
LSID's. That is if the data it represents (the actual bytes) change in any 
way, a new LSID SHOULD be created for the newly formed set of bytes, 
PERHAPS by incrementing the version area in the LSID name (if it is felt 
that is appropriate). 

Other than that the entire LSID URN scheme is nothing more than a shared 
naming convention  (what characters do we assemble in what order to make a 
name string) that assures uniqueness, for anyone to apply to whatever data 
 they see fit and useful to them. It is intended as a convenience to those 
who create and name data, not a burden and is useful as a means to promote 
future interoperability of that data beyond its original system.

Other than uniquely naming of a set of bytes, the LSID is also an anchor 
or handle to metadata documents that describe those bytes, either 
literally or disclosing its relationships to other things described by 
URIs. Metadata documents can come from the people who created any 
particular LSID or from any third party who thinks they want to say 
something about that LSID (both literally and/or to disclose more 
relationships for it). 

Sometimes a LSID names no bytes at all. In this case it is useful as a 
permanent place to "hang" multiple metadata documents. I think of this as 
a conceptual or abstract LSID. This is probably the best form to use if 
one was describing people - after all what one set of bytes describes a 
person?  In the metadata for this "person" LSID, I would described some of 
the literal attributes of the person (age, height?) and include links to 
LSIDs that do have associated bytes for that person. Perhaps to a recent 
photo; the results of a sequence on a tissue sample; a medical report, a 
X-ray image etc. Each of these LSIDs name a particular thing and may 
themselves have associated metadata documents that further describe that 
thing and its relationships. Pretty soon you have an enormous graph of 
interconnections that your software can explore.

In the current specification the data bytes named  by LSID may never 
change but there is no such  restriction on the metadata documents. In 
future versions of the specification we are likely to see metadata split 
into that which changes and that which is always fixed just like the data 
bytes are - this makes  more sense for certain things that describe data - 
for instance the MD5 hash of the data or the date it was  created or 
perhaps the name of the data bytes creator. Future versions of the 
specification are also likely to more closely detail the some of literals 
and relationships software can expect to find in metadata documents, so 
that general software can be written to automatically traverse the growing 
webs of metadata graphs that interconnect LSIDs.

The LSID resolution protocol  was designed to allow client software 1] to 
retrieve a copy of the named bytes if there are any 2] to retrieve copies 
of the metadata documents if there any. 

Happily LSID's are URNs, and therefore a form of URIs, so you can 
certainly use them in RDF, OWL, OWL-S etc. In our systems we happily 
intermix them with the more traditional http style URIs and so far nothing 
bad has happened :-) 

Thats about it.. no magic. 

Kindest regards, Sean

--
Sean Martin
IBM Corp.









Gary Bader <bader@cbio.mskcc.org> 
Sent by: public-semweb-lifesci-request@w3.org
04/07/2005 01:18 AM

To
<Eric.Neumann@sanofi-aventis.com>
cc
<public-semweb-lifesci@w3.org>
Subject
Re: [BioPAX-discuss]  LSID Best practices...







Hi,
This may be a naive question, but does anyone know if LSIDs
could/should be used as RDF IDs for individuals in OWL documents?

Thanks,
Gary

On Apr 6, 2005, at 12:36 PM, <Eric.Neumann@sanofi-aventis.com> wrote:

>
>
> Forwarding this set of best practices proposed for LSID... these may
> be useful to model in the semantic web context.
>
> Eric
>
> Eric Neumann, Ph.D.
>
> Global Head of Knowledge Management
> Sanofi-Aventis Pharmaceuticals
>
>
> -----Original Message-----
> From: biopax-discuss-bounces@biopax.org
> [mailto:biopax-discuss-bounces@biopax.org]On Behalf Of Jeremy Zucker
> Sent: Wednesday, April 06, 2005 2:30 AM
> To: biopax-discuss@biopax.org
> Subject: [BioPAX-discuss] FYI LSID Best practices...
>
>
> http://www-128.ibm.com/developerworks/opensource/library/os-lsidbp/
>
>
>
> _______________________________________________
> BioPAX-discuss mailing list
> BioPAX-discuss@biopax.org
> http://www.biopax.org/mailman/listinfo/biopax-discuss
>
Received on Thursday, 7 April 2005 13:33:49 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:00:40 GMT