HCLSIG BioRDF Subgroup/Tasks/URI Best Practices/Recommendations/PublicResources

From W3C Wiki

[[/../../Recommendations|URI Note main page]]

URIs for Public Resources

Having a global identifier for public resources simplifies data integration when using semantic web technologies (RDF/OWL) because statements about such resources can be trivially merged by their URI. This is true whether the data comes from primary data providers or distributors and also from third party annotators. However, a lack of agreement will result in the proliferation of URIs that will eventually have to be mapped to one another, thereby re-creating the very problem of identifier mapping that we can actively avoid.

Status quo

There are few data providers making their data available in RDF/OWL. Hence, it is unknown what the form of the URI is preferred by these providers. URI directories have already been established (HCLS PURL, Bio2RDF, LSRN) and may serve as a basis to move forward, but these will have to be mapped to one another for integration.

NCBI Resource Locator

The NCBI Resource Locator provides stable, uniform addressing for NCBI web content, making it easy to link to individual records. Some NCBI resources also provide services (like search) through these URLs.

http://view.ncbi.nlm.nih.gov/

Bio2RDF

Bio2RDF publishes their own normalized URIs and maps these to LSIDs, dc:identifier and their Banff Manifesto URN identifier. http://bio2rdf.org/

purl.org/commons/

The PURL solution ([[/../Purls]])aims to provide stable http identifiers whose resolution can be re-located when data providers are so willing. See http://sw.neurocommons.org/2007/uri-explanation.html and http://wiki.neurocommons.org/CommonNaming .

LSID

LSIDs aim to provide both global identifiers with non-http resolution that allows multiple resolvers to register with an authority.

LSRN

LSRN (Life Science Record Name) provides http-based global identifiers for life science resources such that multiple resolvers may be listed. The registry is maintained in RDF (http://www.lsrn.org/lsrn/registry.rdf) and the web site also provides browser based navigation so as to choose alternative data providers.

Handles and DOIs

Pubmed only covers biology and medicine, so another way to identify articles is needed.

Can't get to the metadata from the DOI yet, so not very useful.

What is the correct way to represent a handle as a URI? info:hdl/, and use resolution rules ([[/../URI_Resolution]]) to dereference (when possible)?

info:

The info scheme is to create global identifiers that are location independent and have no resolution mechanism. Precedent: info:pmid/.

Relate via native identifiers and an ontology, not via names

[a public:PubmedRecord ; public:hasId "16228203"]
[a public:Article ; public:hasPmid "16228203"]


New institution

Get a new domain name, buy a hundred-year lease, set up a server, do everything legally and technically possible to set up a sustainable system and garner trust.

Essential Characteristics:

  • Establishes a free and user-managed registry of global identifiers
  • Registry allows any provider to resolve identifiers (i.e. by providing a URL to a RDF/OWL document that says which identifiers it will resolve and how - template regex, content-type)
  • Facilitates mapping (LSID to LSRN to PURL mapping) thereby providing support for alternative identifier/resolution systems
  • Automatically generates HTML/XML/RDF/OWL documents about this identifier for search and agent discovery