Talk:URI-CV Discussion

BioPAX Specification
URIs

all BioPAX URIs should be resolvable HTTP URIs
 * one URI for the most current release of the ontology
 * e.g. http://biopax.org/ontology/biopax.owl
 * one URI for each version of the ontology
 * e.g. http://biopax.org/ontology/level3/biopax.owl
 * ontology versions can be specified as annotations in the OWL ontology block
 * e.g. 'Ontology' '(' [ ontologyIRI [ versionIRI ] ]
 * classes, relations and individuals should use opaque, version-independent URIs
 * e.g. http://biopax.org/resource/1

annotations
 * entities should have human-readable labels using rdfs:label
 * e.g.  rdfs:label "Protein"@en.
 * entities should have human-readable definitions using dc:description
 * e.g.  dc:description "A protein is a molecule composed of amino acids."@en.

versioning
 * use owl versioning mechanisms in the ontology header
 * indicate version increment with owl:priorVersion
 * indicate backwards compatibility with owl:backwardCompatibleWith
 * indicate incompatibility with owl:incompatibleWith
 * create new URIs for new classes, relations, individuals

BioPAX Data
using BioPAX ontology
 * data providers create a new document (with an ontology URI in its domain) that imports biopax ontology
 * e.g.  owl:imports 
 * data providers create new entity URIs in one of two ways:
 * for redirection e.g. http://purl.org/biopax/reactome/resource/1
 * for direct resolution e.g. http://reactome.org/biopax/resource/1
 * reference to external vocabularies will use a common URI from which other URIs can be derived
 * the minimal information is the dataset and the identifier.
 * the dataset term can be drawn from a dataset registry

there are several of these including
 * http://semanticscience.org/projects/registry/ (used by Bio2RDF)
 * http://lsrn.org (used by SADI; to be replaced by the semanticscience registry)
 * http://neuinfo.org/nif/nifgwt.html?registry=yes (in discussions to map with semanticscience registry)
 * database citation controlled vocabulary (MI:0444) (has been integrated to the Miriam)
 * http://www.ebi.ac.uk/miriam

-> this points to a potential collaborative effort to construct a registry that is easily managed - and can be used as a source for generating all equivalent URIs --Michel 10:49, 18 February 2011 (EST)
 * e.g. http://bio2rdf.org/go:0000001
 * e.g. http://purl.org/obo/owl/GO#GO_0000001
 * e.g. urn:miriam:obo.go:GO%3A0000001

So far, above - I fixed lists, the Miriam URN, and also added about the MI "database citation" CV...

--Rodche 17:33, 18 February 2011 (UTC)

I think classes and relations shouldn't have semantic opaque URL. This may be correct, by very counter-intuitive for many not "hard core" users. Especially for relations which are high level, you want to read: obo.org/part-of (or the like), rather rel0001 + rdfs:label

--Andrea 17:45, 18 February 2011 (UTC)

A different thing is for instances/resources. If we provide a scheme, that has to be opaque.

test

ContolledVocabulary to be Xref
Can we remove 'xref' froperty from ContolledVocabulary (thus - all its sub-classes) and make it a sub-class of Xref? Why not? --Rodche 16:11, 18 March 2011 (UTC)