W3C home > Mailing lists > Public > public-hcls-coi@w3.org > January to March 2008

Re: An argument for bridging information models and ontologies at the syntactic level

From: Xiaoshu Wang <wangxiao@musc.edu>
Date: Wed, 26 Mar 2008 17:39:42 +0000
Message-ID: <47EA8A5E.9090602@musc.edu>
To: "Ogbuji, Chimezie" <OGBUJIC@ccf.org>
CC: public-hcls-coi@w3.org, public-semweb-lifesci@w3.org

Chimezie,

Ogbuji, Chimezie wrote:
>
> For some time I have had a concern about a theme in the more common 
> approaches to bridging 
> information models and ontologies as a path towards bringing the 
> advantages of the Semantic Web technologies to 'legacy' healthcare 
> terminology systems.
>
A good starting point.  However, here we must have to know the nature of 
"information model" before addressing their bridges.  If the model is a 
DB system, they are not too far apart from RDF/OWL.  But if they are an 
OOP model, they are very far apart.  Because a programming object is 
about what an object does but what an object is, which is address by 
ontology.  They are orthogonal to each other.  But both DB, OOP and XML 
are essentially closed world, hence, there cannot have a default binding 
scheme's like object-to-XML mapping.  I have discussed my viewpoint more 
in here. http://dfdf.inesc-id.pt/tr/orb.
>
> To get right to the point, 1) I consider approaches that attempt to 
> perform this bridging directly between information models and 
> ontologies as examples of this 'anti-pattern.' 2) I think that 
> performing this bridging at the syntactic level addresses the 
> important problem of properly separating these two  in a way that 
> emphasizes their strengths.
>
The earlier analysis showed that the "anti-pattern" is natural and 
correct. :-)

The syntactic level is also correct in my opinion, but I have described 
it differently.  A resource (or entity) denoted by a URI carries two 
kinds of meanings.  One is the extensional meaning.  For instance, me - 
the person - is what it is.  The second one is the intensional 
meanings.  Hence, I can write a document, either casually or RDF 
document about me.  In the former, I will use natural language and in 
the latter I will put in a lot of logical constrains etc.  But as all 
reality, nothing can be completely identified by its extensional 
meaning.  The latter changes (i.e., RDF etc will change) while the 
former doesn't.  But the change in the former may lead to logical 
inconsistencies, hence the issue is how to keep a meaningful 
relationship between them so that our conceptualization about an entity 
can advance while their extensional identity remains stable. 
>
> In a perfect world, a messaging dialect (such as HL7 RIM or even Atom 
> for that matter) would be developed with a formal conceptualization as 
> part of its specification.  This conceptualization would be captured 
> in a formal knowledge representation (such as some particular fragment 
> of FOL, for instance) as a way to reach consensus on the 'real world' 
> entities that the messages refer to.
>
The (my) solution is to develop all those useful ontologies, such as 
HL7, GO etc. in http://dfdf.inesc-id.pt/ont/o3#Vocabulary.  A vocabulary 
is a URI whose RDF document has no logical constraints.  All intensional 
meaning is put in separate URIs.  When an application needs to use them, 
they can simply import which ever version of the intensional meaning for 
their use.  This is the only way I think that can gradually advance our 
conceptualization about certain reality while keeping the URI for the 
reality a constant.   Honestly, I think both RDF and OWL should follow 
the same principle.  OWL1.0 and OWL1.1 won't cause any compatibility issues.

I have written a manuscript to HCLS-08 at Beijing, where I used BioPAX 
as an example to show how its current practice prevents it from making 
their URI stable.  But the manuscript was turned down with the reason 
that "the example is too simple." The example that I used is the 
biopax:NAME.  Honestly, I don't know what is the logic behind that.  If 
a system cannot even get the simplest conceptual thing such as NAME 
right, what else can it get right? 

(On a side note, if it is the reviewer who is reading this, please don't 
take it personally.  Me writing this example is to show Chimezie how 
difficult he may get his idea crossed. )

Most existing ontology effort has never considered the possibility of 
change and sharing because their conceptual model is still in the mode 
of DB, OOP, and XML.  They all look similar to RDF/OWL but they are very 
very different.  Now, we only see big monolithic ontology, which is very 
hard to change and evolve.  But things always change!  Our software 
engineer experience has showed us how we must have to manage change by 
carefully removing the code-dependencies.  There is no exception in 
ontology development.  We must develop our ontology with dependence and 
change in mind.  Sooner or later, people will realize this.  I hope it 
is sooner....

Xiaoshu
Received on Wednesday, 26 March 2008 17:40:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 26 March 2008 17:40:52 GMT