RDF - WTF?
Notes by Justin Miller – sorry for typos/omissions
RDF is machine-readable information about information triples: subject-predicate-object anything in a relational DB can be represented in triples
have a username: “joesmith is a user of X”
SPARQL: query language across resources predicates themselves can, as human knowledge, be described in RDF as well alternatives to using XML for RDF (which turned out to be verbose and bad)
Turtle
practical applications
biggest: FOAF (friend of a friend)
describes social network information
build it yourself, publish it on your site
becomes your social network identity
LiveJournal auto-publishes it for everyone
WordPress, Blogger, Movable Type as well for years
vast majority of live RDF today is FOAF
post info about RDF in Drupal
http://drupal.org/project/rdf http://drupal.org/project/fileframework
tool called Zotero - like Evernote
Firefox plugin
bookmark a site, tag it (like Delicious)
also: senses bibliographic info on the page & extracts it for your records
example: your on an Amazon book page, Zotero saves info into your database
question: why use FOAF vs. some other format?
it’s already a standard tools already exist
N3 allows you to define formalae
example: X is father of Y, A is cousin of Y, implies X is uncle of A
Thetus: “knowledge discovery software” (based in Portland) interesting application: trust networks as extension of FOAF downside: hard to store a lot of it
generally you store it in your own format and export RDF (or make it available on demand) Billion Triple Challenge: project trying to tackle this challenge
question: anyone looked into using distributed things like CouchDB to store it?
problem is they are usually key-value, not triples someone needs to build a distributed storage DB for this? storage of quads: adding security or lineage info as well
DBpedia: http://en.wikipedia.org/wiki/DBpedia
trying to take all of Wikipedia’s info and putting it into RDF
question: can we mashup the CIA World Factbook with Wikipedia?
that’s the dream of RDF, really if everything has an absolute URI, then it’s more easily possible
question: what about connecting FOAF info between, say, LJ and Blogger?
unfortunately a lot of current use is internal to organizations but you could write your own FOAF integrating across services and publish that
question: what happens when things change names (say: countries)
depends on how you define your predicates
“this country is another name for this country”
question: can you explain the OWL (Web Ontology Language) language?
http://www.w3.org/TR/owl-features/ includes mathematical and logical relationships example: saying a pine tree is a type of plant implies that plants include trees
also check out Yahoo’s YQL session today at 6:00pm