Bundles/medline/abstracts-index

NOTE: There is no longer a separate bundle for indexing Medline; the indexing is done by a script that belongs to the abstracts bundle.

A tiny script (see below) creates a full text index of the medline/abstracts bundle.

To enable full text indexing in Virtuoso: DB.DBA.RDF_OBJ_FT_RULE_ADD (null, null, 'comment'); This can be limited to a single graph or even a single property: DB.DBA.RDF_OBJ_FT_RULE_ADD('http://purl.org/science/graph/medline/abstracts',                           'http://purl.org/dc/terms/abstract',                            'medline abstracts') This command took 14 hours to execute (June 2009, development instance). Yrjänä Rankka of Virtuoso writes: "It is easier (faster) if this is done prior to loading the data." This seems to be true; loading the abstracts bundle, including indexing, only took 1.5 hours (October 2009, beta instance), and indexing was turned on as above before loading the abstracts.

Full text indexing resulted (June 2009) in a 7.8 G increase in the size of the DB.DBA.RDF_OBJ_RO_DIGEST_WORDS table.

Alan Bawden writes: "All I know about this stuff comes from , which is a page I bookmarked as potentially interesting sometime last year, but which is no longer pointed to by the rest of their documentation."

(Of course, you can do this query at the NCBI web site, but the above can be combined with other conditions.)

See also Bundles/medline/abstracts