Wednesday, 6 June 2007

Banff Manifesto

From Bio2RDF, I stumbled across this link to the Banff Manifesto. To quote:
Some participant from the HCLS-DI2007 workshop, and from the I3 Workshop propose to the community of life science to establish an authority for attributing namespaces used in semantic web to build URIs. This authority will also establish a set of rules for the construction of well formed RDF documents. When someone will use URI approved from the BM authority, they will know that some standard of naming are upheld. The set of rules are good pratices that need to be followed to be in recognized by the BM community.

Much in the way DOI work, BM will certified the appartenance of a namespace to a data provider and disambiguate the naming for it.

Essentially, it proposes that identifiers are written as


where the authority_namespace is bm.Resolution (at least in the examples presented) is via a persistent URL, e.g. bm:uniprot:p26838 becomes

This looks to be much the same idea as bioGUID. Interestingly, one of the exmaples of the URI chaos at present is a DOI for a Protein Data Bank record (doi:10.2210/pdb2b3y/pdb. This DOI resolves to a compressed file served by FTP. You can see the web version here. The CrossRef XML for this DOI is interesting, in that sort-of treats the record as an article, but also has a component-list (which curiously states that this record is a part of itself).