The medline/abstracts bundle provides data from MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine. It consists of a set of triples, one for each MEDLINE record that contains an abstract, asserting that the article (the subject of the MEDLINE record) has that abstract.

Sample query:

select ?s ?o
  graph <http://purl.org/science/graph/medline/abstracts> {
    ?s <http://purl.org/dc/terms/abstract> ?o .
 } limit 5

(I probably used the wrong Dublin Core URI here for the medline/abstracts bundle. It appears there are at least three sets floating around.)

Some pragmatic information on this bundle (as of June 2009):

  • consumes 8.8G of disk (compressed) in RDF
  • took 71 minutes to load into Virtuoso 5.0.11
  • when loaded resulted in a 1.3G increase in the size of the DB.DBA.RDF_OBJ table (and smaller increases in a few other tables)

See also Bundles/medline/abstracts-index