Differences between v0 and v1


 * Version 0 = Banff demo system (see RDF graphs v0, query form)
 * Version 1 = September 2008 Neurocommons RDF distribution

Incompatible content changes

 * Almost all of the named graph URIs are different now, so in the event you need to refer to one of these graphs, check the name first. Consult individual bundle documentation linked from Bundles page for graph name.  However, you shouldn't need to use FROM or WHERE any more as default-graph performance has improved significantly. In general it should work to just remove all FROM and WHERE directives.
 * In Medline years/titles, year of publication has datatype xsd:gYear. In version 0 it was untyped string.
 * The URI for the part_of relation has been canonicalized to http://www.obofoundry.org/ro/ro.owl#part_of . Formerly it was http://purl.org/obo/owl/obo#part_of or http://purl.org/obo/owl/part_of; in the current (August 2008) OBO downloads it is http://purl.org/obo/owl/OBO_REL#part_of.

Compatible content changes

 * We have updated the content to current versions of MeSH, Medline, all OBO ontologies, GOA, Homologene, and other sources.
 * We have added nine additional species to our NCBI extractions. (TBD: list them somewhere.)
 * The default graph contains the union of all of the other graphs, and queries that do not specify a graph perform better now than formerly. There is no need for FROM or GRAPH clauses, and if you have them in your queries you should remove them.

SPARQL endpoint changes

 * It appears that due to some change in Virtuoso, it is now necessary to write, in filters, str(?foo) where a simple ?foo formerly sufficed.
 * It appears that in older versions of Virtuoso, untyped literals were given a type of xsd:string. Now this doesn't happen; instead untyped literals have no specified type in query results. (This difference shows up in the XML serialization of sparql query results.)