10M RDF Triples

A conversion of UMBC ~10M RDF triples from the original SQL dump
from UMBC including some fixes and filtering to give legal syntax.

Outputs:
  triples.nt.bz2    N-Triples version - 10,429,951 triples

  triples.rdf.bz2   RDF/XML version   - 10,429,947 triples

Processing Notes

I started from the data described at
http://ebiquity.umbc.edu/resource/html/id/126
in file http://ebiquity.umbc.edu/share/triples.sql.gz
which is a SQL database dump (rather useless in itself):
  -rw-r--r--  1 dajobe dajobe 1799342725 Jun 16 13:44 triples.sql

I loaded it into a mysql database and wrote a custom perl DBI
script mysql2ntriples to turn it into legal N-Triples.

It generated the N-Triples (updated 2006-04-14 to fix bad \u hex)
  -rw-r--r-- 1 dajobe dajobe 1625025120 Apr 14 11:59 triples.nt
which via rapper (see Makefile) generated the RDF/XML:
  -rw-r--r-- 1 dajobe dajobe 2117989148 Mar 14 12:04 triples.rdf

However this has triples that cannot be written in XML such as ones
refering to ASCII 7 in literals; which isn't allowed in XML 1.0
character data.

It also discards some triples with predicates that cannot be written in
RDF/XML such as
  http://example.org/#
  http://www.w3.org/2000/10/swap/pim/doc.n3#@@
  http://example.org/property/
 
Only the syntax is valid, there are also invented RDF namespace terms
such as:
  rdf:date rdf:testCsae rdf:asserts rdf:range rdf:domain rdf:dtype

See Makefile for the actual operation for making triples.rdf from
triples.nt using rapper (from http://librdf.org/raptor/)

Dave Beckett
http://purl.org/net/dajobe/
2006-03-14