RE: Antwort: RE: Semantic web article in Nature Biotechnology

-Wafik,

> 1. In a Profile - how can it merge ontologies without having 
> "concepts"
> under its namespace -- what good will this bring us??  When you say 
> "merge"

For instance, I have a 2D Gel ontology (CC2GO)
(http://www.charlestoncore.org/ontology/2005/08/gel#) and a BOSS.  Now I
want to make the cc2go:Gel a boss:Data but don't want to tie them up because
someone may like my CC2GO but not BOSS, what am I going to do?  The
statement, cc2go:Gel a boss:Data must be made at some point, making it at
ontology level is not ideal due to its strong dependenccy. So, you create an
o3:Profile and manage your own application profile without forcing others to
buy into yours.

> I assume merge and not integrate.  For example, A "protein_name" and 
> "protein_length" can be in one ontology and another ontology has 
> "protein_state" and "protein_distance".  Now merging is putting all 
> four attributes in a name_space but doesn't create a dependency 
> between the attributes -- correct?  This means if I have a protein 
> name I will not be able to know its state.  If yes, than what is the 
> benefit of merging them?

I think merge is the same as integrete.  If conceptualization overlap, you
have to align them.  See an example profile here,
http://www.charlestoncore.org/profile/2005/08/gel#.  So, the second benefit
of Profile is instead of "merge" it at data instance level, you manage them
at Profile.  If someone create a Profile that you like, you use them without
making you to go through all the trouble again.

> 2. Complex ontologies should be "normalized" into local ontologies.  
> Hmm, ..
> How do you normalize an ontology and why?  

You create a new namespace for profile, move all the "merging" or "aligning"
statement in Profile, that leaves you a clean local ontology.

Xiaoshu

Received on Wednesday, 12 October 2005 20:40:02 UTC