On 23/2/09 20:11, Frederick Giasson wrote:
Hi all,
David, I know you already know my personal view of this stuff with our
recent private interactions & on the umbel mailing list. For the LOD
mailing list readers, here is the background for this conversation:
http://groups.google.com/group/umbel-ontology/browse_thread/thread/6bd977e50e1f372b
Matthias suggested to used rdfs:seeAlso. I think it would make sense to
createa subproperty of this property for that kind of purpose.
In UMBEL we created a property that "is like" owl:sameAs, but that
doesn't entail as much as sameAs does. *I am not suggesting* to use it
for this specific use case, but *I would* suggest it to use it instead
of owl:sameAs when "automatic" linkage is done between two datasets
where issues can arise. This property is called "umbel:isLike" and come
from the UMBEL ontology here:
http://umbel.org/ontology/umbel.n3
The umbel:isLike property is described as:
"he property umbel:isLike is used to assert an associative link between
similar individuals who may or may not be identical, but are believed to
be so.
Certainly a useful concept to have name(s) for, ... but I think not
quite what is needed to link these specific datasets. The W3C wordnet
data is a collection of descriptions of linguistic concepts. The OpenCyc
things they're linked to are not linguistic entities, but the real world
things the words are associated with. So they're not near-identical
similar objects, but things that are generally from disjoint classes.
Paris the place, vs "Paris" the word (or the set of synonymous words
_for_ Paris...). I think this is a distinction worth keeping reasonably
explicit.
cheers,
Dan
