xrefs type and labeling about bio2rdf-scripts HOT 11 CLOSED

So ":a" is the dataset URI?

from bio2rdf-scripts.

a data item uri

from bio2rdf-scripts.

another option was adding the dataset namespace via dc:source
ns:id dc:source "namespace" ;

from bio2rdf-scripts.

Could you give an example how this would work for a concrete resource/dataset combination (say uniprot for example)? I am having trouble visualising it.

from bio2rdf-scripts.

refseq:54321 refseq_vocabulary:x-uniprot uniprot:Q12345 .
uniprot:Q12345
dc:identifer "Q12345" ; // or dc:identifier "uniprot:Q12345"
dc:source "uniprot" // or dc:source http://bio2rdf.org/uniprot
a uniprot_vocabulary:Resource

the value in having the dc literals in the above form is that they are indexed and can be used to construct a URI

from bio2rdf-scripts.

Okay, much clearer.

Up until recently dc:identifier was always "ns:id", so I would personally prefer, for backwards compatibility all the way back to the original Banff Manifesto, if we stayed with that convention. We could use another predicate to have just the "id" available as a literal.

dc:source is specified to provide a link from a resource to a related resource that the given resource is derived from. The specification recommends that a formal identification scheme be used, so "uniprot" isn't the clearest identification that we could have there. I would also prefer http://bio2rdf.org/dataset:uniprot or http://bio2rdf.org/dataset/uniprot or something similar to that, rather than http://bio2rdf.org/uniprot

http://dublincore.org/documents/2012/06/14/dcmi-terms/?v=terms#terms-source

from bio2rdf-scripts.

i don't know of anything that matches exactly then. we could do something
like

bio2rdf_vocabulary:source-identifier "Q12345"
bio2rdf_vocabulary:source-dataset "uniprot"

On Mon, Mar 25, 2013 at 11:27 PM, Peter Ansell [email protected]:

Okay, much clearer.

Up until recently dc:identifier as always "ns:id", so I would personally
prefer, for backwards compatibility all the way back to the original Banff
Manifesto, if we stayed with that convention. We could use another
predicate to have just the "id" available as a literal.

dc:source is specified to provide a link from a resource to a related
resource that the given resource is derived from. The specification
recommends that a formal identification scheme be used, so "uniprot" isn't
the clearest identification that we could have there. I would also prefer
http://bio2rdf.org/dataset:uniprot or http://bio2rdf.org/dataset/uniprotor something similar to that, rather than
http://bio2rdf.org/uniprot

http://dublincore.org/documents/2012/06/14/dcmi-terms/?v=terms#terms-source

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/198#issuecomment-15439042
.

Michel Dumontier
Associate Professor of Bioinformatics, Carleton University
Chair, W3C Semantic Web for Health Care and the Life Sciences Interest Group
http://dumontierlab.com

from bio2rdf-scripts.

Would the following be a useful compromise?

dcterms:identifier "uniprot:Q12345"
bio2rdf_vocabulary:source-identifier "Q12345" 
bio2rdf_vocabulary:source-dataset "uniprot"
bio2rdf_vocabulary:source-dataset-uri <http://bio2rdf.org/dataset:uniprot>

Should we also recommend the usage of rdfs:label where possible to give a human readable label if we can retrieve one? This would make scripts a lot slower but it may be useful enough to justify the additional requests.

rdfs:label "Ino eighty subunit 3"

from bio2rdf-scripts.

On Mon, Mar 25, 2013 at 11:50 PM, Peter Ansell [email protected]:

Would the following be a useful compromise?

dcterms:identifier "uniprot:Q12345"
bio2rdf_vocabulary:source-identifier "Q12345"
bio2rdf_vocabulary:source-dataset "uniprot"
bio2rdf_vocabulary:source-dataset-uri http://bio2rdf.org/dataset:uniprot

the alternative here is to put
a uniprot_vocabulary:Resource.

and when RDFizing the resource we'll see void:inDataset

Should we also recommend the usage of rdfs:label where possible to give a

human readable label if we can retrieve one? This would make scripts a lot
slower but it may be useful enough to justify the additional requests.

rdfs:label "Ino eighty subunit 3"

this is questionable - should we generate a label when we're not processing
the dataset in which it is described? The problem is if the source dataset
changes to the label, we'll see multiple values if we merge the graphs.
That being said, sometimes that dataset does provide us with labels that
might be otherwise hard to get.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/198#issuecomment-15439603
.

Michel Dumontier
Associate Professor of Bioinformatics, Carleton University
Chair, W3C Semantic Web for Health Care and the Life Sciences Interest Group
http://dumontierlab.com

from bio2rdf-scripts.

ok, so let us allow the assertion of a label for an xref identifer provided by a third-party as an rdfs:label "text [ns:id]", but also mandate dc:title "text" in the primary dataset. that way if there is an inconsistency in label, we can fall back to dc:title as the dataset-specific one.

from bio2rdf-scripts.

Sounds good to me.

from bio2rdf-scripts.