The RDF Working Group at W3C has been busy with an update to the RDF standards, and this work is now nearing its end. Here I summarise what's new, what's been controversial, and also give a very brief intro to the RDF data model.
Richard Cyganiak — DERI Friday Talk, 26 July 2013
WHAT'S NEW IN RDF 1.1?
World Wide Web Consortium (W3C)
• Standards body responsible for the Web
• DERI is a member
• What W3C can do for you:
– Community Groups
– Work with experts, network, see how the sausage
• RDF — A quick overview
• What's new in RDF 1.1?
• The controversies and stuff that didn't make it
Foundation for a family of standards
SPARQL: Querying RDF data
RDF Schema: Documenting the meaning of RDF data
OWL: Formalizing the meaning of RDF data
RDF/XML: Writing RDF data in XML
RDFa: Embed RDF data in HTML
R2RML: Mapping relational data to RDF
GRDDL: Mapping XML data to RDF
DC, Org, FOAF, SIOC, DCAT, VoID, …:
Describe particular domains as RDF data
• First update since 2004
• RDF-WG started in 2011, expected to end
• Chartered to do:
– Fixing some well-known issues
– "Paving the cowpaths"
• Nothing new in RDF Schema and RDF/XML
• Adopted from SPARQL — "Named Graphs"
• Two data structures:
– RDF graph
– RDF dataset
• Multiple RDF graphs, each named with a URI
• A.k.a “quads”
• Unnamed “default graph”
– can be used for metadata
integration, provenance, versioning, …
• xsd:duration, xsd:dayTimeDuration, xsd:yearMonthDuration
– Unlike xsd:dateTime, the time zone is not optional
– "Hello"@en now has a datatype
– Problem: No language tag!
– No longer needs to be canonical XML
– "xxx" and "xxx"^^xsd:string are now 100% the same thing
– Now SPARQL compatible
– Turtle with named graphs
– WG Note
– WG Note
and stuff that didn't make it
Deprecating some of the bad stuff
rdf:Alt, rdf:Bag, rdf:Seq
For every bad feature, there is a vocal user
who can't live without it
Literals as subjects
• 1 math:lessThan 2
• "2013"^^xsd:gYear time:isBefore "2014"^^xsd:gYear
• "Galway" ex:isTitleOf<http://dbpedia.org/resource/Galway>
• Most specs and implementations could be
easily adapted to allow this
• Not worth it
Semantics of RDF datasets
• What logical statements does an RDF dataset
– How can we reason over them?
• In a named graph, what is the relationship
between the graph name URI and the RDF graph?
– Same as? Is described by? Contains?
• No solution fits all use cases. It's left unspecified.
– No standard way to reason over RDF datasets
Nested and anonymous graphs
• What about: "Bob said that Alice said XXX"?
• Should we allow graphs as a new type of node
in RDF triples?
– URIs, literals, blank nodes, RDF graphs
• Little implementation experience (but N3!)
• Requires new DBs, different query language