- From: Matt <matt.halstead@auckland.ac.nz>
- Date: Wed, 13 Sep 2006 12:29:33 +1200
- To: AJ Chen <canovaj@gmail.com>
- Cc: "Ivan Herman" <ivan@w3.org>, public-semweb-lifesci@w3.org
- Message-Id: <7E343BD1-D2CC-40BD-803F-BEEC590977BD@auckland.ac.nz>
It's probably quite important to define various relation classes for the aggregated properties we tend to relate to a person. I would imagine this comes under standard upper ontologies. It would necessarily need to include definitions of FOAF and vCard so that we could classify across current data records. I doubt though that much more than this would be designed from the top down since it tends to be the evolution of current application data formats that will govern the actual graph structures used to represent and transport the data. So I would opt for an attempt to identify the current players in terms of each other using an upper ontology rather than try to force a single design on everybody. Matt On 13/09/2006, at 12:02 PM, AJ Chen wrote: > In developing SPE ontology, I have tried to re-use FOAF and vCard, > but unfortunately found little can can be re-used. One main reason > is that, although they may have the terms, the definitions of these > terms usually don't match what's required by the Person class in > SPE ontology. The problem mostly comes from the "range" of a > ObjectProperty or DataProperty. > > I wish there was a Person class defined in RDF or OWL that can be > re-used in any application and easily extended to include special > properties in specific domains. Hope the next attempt by W3C will > create just that. > > AJ > > On 9/12/06, kei cheung <kei.cheung@yale.edu> wrote: > > Hi Ivan et al., > > Based on my limited experience, a person in the life science and > healthcare context can be considered as a subject or patient (which > can > be a subclass of person). Of course, there are other roles a person > can > play (e.g., doctors, researchers, and authors). For genetic studies, a > group of subjects/indviduals may be a family/pedigree. In this case, > relationships among these family members may include Father_of, > Mother_of, Child_of, etc. Other types of relationships can be inferred > (e.g., uncle, sibling, etc). For popualtion genetics, we need to know, > for example, the ethnicity of the subjects and the geographical > information about the population to which the subjects belong. > There can > be mutliple types of ID's (e.g., patient id, cell line id, etc) > associated with a person (whether the person is a subject or patient). > Sometimes a dummy person (not a real person) is needed to fill in the > missing data ( e.g., in linkage data analysis). I am not exactly clear > how these specific HCLS use cases of persons would impact the generic > modeling of person. Maybe this is something we all need to think more > about. This is just my 2-cent thought. > > Best, > > -Kei > > > Ivan Herman wrote: > > >Dear all, > > > >we would need some feedback... > > > >There were some brainstorming on what vocabularies to use for the > simple > >notion of 'Person' in various settings. There is old W3C note for > an RDF > >version of vCard[1], but another version was created by Norm Walsh a > >while ago[2]. And, of course, there is FOAF. > > > >The issue came up because some people would like us to update the old > >[1] note but, if we want to do that seriously, it is not necessarily > >that easy (the vCard spec itself is not soooo o.k.). > > > >Hence the question as a feedback: what does the HCLS community use > for > >something like 'Person'? > > > >Thanks for the feedback > > > >Ivan > > > > > >[1] http://www.w3.org/TR/vcard-rdf > >[2] http://norman.walsh.name/2005/12/12/vcard > > > > > > > > > > > -- > AJ Chen, PhD > http://web2express.org
Received on Wednesday, 13 September 2006 00:30:12 UTC