Construct RDF triples from the Europeana SPARQL endpoint on different AV resource types both enriched with GeoNameIDs and based on the name of the municipality

gistfile1.r

#######################################################################
#
# Construct RDF triples to an in-memory triple store
# from the Europeana SPARQL endpoint on
# different AV resource types a) enriched with GeoNameIDs and 
# b) based on the name of the municipality.
#
# Save triple store as RDF/XML.
#
# For getting the names of the municipalities (object comms and Ncomms below), 
# and querying DBpedia on GeoNameIDs (object geo.df below), see 
# https://gist.github.com/tts/5237811
#
# More on this in 
# http://tts2.blogspot.fi/2013/03/some-europeana-av-resources-related-to.html
#
# Tuija Sonkkila 25.3.2013
#
#######################################################################

library(rrdf)

eu_endpoint <- "http://europeana.ontotext.com/sparql"

bigstoreNames <- new.rdf()

for (i in 1:Ncomms) {
  
  store <- new.rdf()
  
  c <- comms[i]
  
  cat("Next constructing ", c, " (", i , "/", Ncomms , ")\n", sep = "")
  
  store <- construct.remote(eu_endpoint, 
                            paste("CONSTRUCT { ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                                         <http://purl.org/dc/elements/1.1/title> ?title ;
                                  <http://purl.org/dc/elements/1.1/spatial> '", c, "' . }
                                  WHERE {
{
{
                                  SELECT ?resource ?title ?type WHERE {
                                  ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                  <http://purl.org/dc/elements/1.1/language> 'fi' ;
                                  <http://purl.org/dc/elements/1.1/title> ?title ;
                                  <http://purl.org/dc/elements/1.1/description> ?d .
                                  FILTER ( regex(?d, '", c, "') )
                                  FILTER ( ?type != 'TEXT' )
                                  }
}
}
                                  UNION
{
{
                                  SELECT ?resource ?title ?type WHERE {
                                  ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                  <http://purl.org/dc/elements/1.1/language> 'fi' ;
                                  <http://purl.org/dc/elements/1.1/title> ?title ;
                                  <http://purl.org/dc/elements/1.1/subject> ?s .
                                  FILTER ( regex(?s, '", c, "') )
                                  FILTER ( ?type != 'TEXT' )
                                  }
}
                                  UNION
{
{
                                  SELECT ?resource ?title ?type WHERE {
                                  ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                  <http://purl.org/dc/elements/1.1/title> ?title ;
                                  <http://purl.org/dc/elements/1.1/language> 'fi' ;
                                  <http://purl.org/dc/terms/alternative> ?a .
                                  FILTER ( regex(?a, '", c, "') )
                                  FILTER ( ?type != 'TEXT' )
                                  }
}               
}
                                  UNION
{
{
                                  SELECT ?resource ?title ?type WHERE {
                                  ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                  <http://purl.org/dc/elements/1.1/language> 'fi' ;
                                  <http://purl.org/dc/elements/1.1/title> ?title .
                                  FILTER ( regex(?title, '", c, "') )
                                  FILTER ( ?type != 'TEXT' )
                                  }
}               
}
                                  UNION
{
{
                                  SELECT ?resource ?title ?type WHERE {
                                  ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                  <http://purl.org/dc/elements/1.1/language> 'fi' ;
                                  <http://purl.org/dc/elements/1.1/title> ?title ;
                                  <http://purl.org/dc/terms/spatial> '", c, "' .
                                  FILTER ( ?type != 'TEXT' )
                                  }
}
}
}
                                  }", sep = ""))

                           
  
  bigstoreNames <- combine.rdf(store, bigstoreNames)  
  
}

#####################################
#
# CONSTRUCT triples from Europeana 
# by the GeoNameID of the municipality
#
#####################################

geoids <- geo.df[geo.df$id != 0, ]

# Excluding Rauma because it generates lots of Norwegian false positives
geoids <- geoids[geoids$c != 'Rauma',]

Ngeoid <- nrow(geoids)

bigstoreGeoids <- new.rdf()

for (i in 1:Ngeoid) {
  
  store <- new.rdf()
  
  c <- geoids$c[i]
  
  cat("Next constructing ", c, " (", i , "/", Ngeoid , ")\n", sep = "")
  
  store <- construct.remote(eu_endpoint, 
                            paste("CONSTRUCT { ?resource <http://www.europeana.eu/schemas/edm/type> ?type ;
                                                         <http://purl.org/dc/elements/1.1/title> ?title ;
                                                         <http://purl.org/dc/elements/1.1/spatial> '", c, "' . }
                                  WHERE {
                                    ?euProxy <http://www.openarchives.org/ore/terms/proxyFor> ?resource;
                                             <http://www.europeana.eu/schemas/edm/hasMet> ", geoid, " ;
                                             <http://www.openarchives.org/ore/terms/proxyIn> ?euAggr .
                                    ?euAggr <http://www.openarchives.org/ore/terms/aggregates> ?providerAggr .
                                    ?providerProxy <http://www.openarchives.org/ore/terms/proxyIn> ?providerAggr ;
                                                 <http://purl.org/dc/elements/1.1/title> ?title ;
                                                 <http://www.europeana.eu/schemas/edm/type> ?type .
                                  FILTER ( ?type != 'TEXT' )
                                  }", sep = ""))
  

  bigstoreGeoids <- combine.rdf(store, bigstoreGeoids)  
  
}

bigstoreAll <- combine.rdf(bigstoreNames, bigstoreGeoids)

# summarize.rdf(bigstoreAll)

save.rdf(bigstoreAll, "europeana_rdf_store.xml", format = "RDF/XML")

# Convert from RDF/XML to Turtle/N3 with Jena rdfcat
# http://jena.apache.org/documentation/javadoc/jena/jena/rdfcat.html
#
# $JENAROOT/bin/rdfcat -out ttl -in RDF/XML -x europeana_rdf_store.xml >europeana_rdf_store.ttl