BM Thesauri

Name	Size	Creator	Creation Date	Comment
File BM-thesauri.rar	20.77 MB	Vladimir Alexiev	Jan 10, 2013 18:01

Some of the BM thesauri are shown below. All are described in Meta-Thesaurus and FR Names#Meta-thesaurus table

Thesaurus	Terms
Escape	88
Ethname	3333
MatCult	1440
Material	2010
Object	5802
Places	45521
School	475
Subject	1429
Technique	594
Treatment	719
Ware	659

Counts include both authorised and unauthorised terms. Unauthorised terms are not in the main hierarchy so they don't carry so much semantic content although they may have scope notes.

Obsolete

Name	Size	Creator	Creation Date	Comment
ZIP Archive thesaurus-ttl.zip	788 kB	Vladimir Alexiev	Jul 20, 2011 09:01	material, object, place in Turtle
ZIP Archive thesaurus-merlin-csv.zip	1.07 MB	Vladimir Alexiev	Jul 20, 2011 09:01	material, object, place in Merlin TTL
File bm-csv2ttl.pl	2 kB	Vladimir Alexiev	Jul 20, 2011 09:01	convert from CSV to Turtle
XML File thesaurus-place-china-merlin.xml	1.74 MB	Vladimir Alexiev	Jul 18, 2011 12:37	XML (Merlin): places in China
File thesaurus-place-china.rdf	5 kB	Vladimir Alexiev	Jul 18, 2011 12:37	RDF: places in China

Conversion to SKOS

Vlado wrote a simple Perl script bm-csv2ttl.pl to convert BM thesauri from Merlin CSV to RDF Turtle in SKOS.
Each term maps to a simple SKOS fragment (places have some extra info):

The mapping is described in the next section

Merlin CSV format

Comma-separated. Last two fields may include \n and \r and are enclosed in ". May include full UTF8 (eg Chinese hieroglyphs).

Fields (prefix (bm|mus)auth_thes):

name: thesaurus type, eg OBJECT, MATERIAL, Place. Maps to lowercase:
skos:inScheme id:thesauri/\L$name\E ;
id: code, eg x6202. Maps to the row's URI
id:thesauri/$id
term: label. Maps to
skos:prefLabel "$term" ;
term_discrim: discriminator
- if 'deleted': skip row
- else adds a different sense to term
  - eg: "yoke (harness)" vs "yoke (clothing)" vs just "yoke" (which is a load-carrier).
  - eg "Messina (Transvaal)" vs "Messina (province - Sicily)" vs "Messina (city - Sicily)"
- References use this syntax (eg "burin (stone tool)")
name_type (for Place only): 'modern', 'archaic'. Maps to
crm:PX.place_name_type "$name_type" ;
use_for: synonyms. Semicolon-separated, maps to multiple triples:
crm:PX.use_for $use_for ;
r_terms: related terms. Not used for now
b_term: broader terms that create the hierarchy
- Poly-hierarchical (i.e. multiple inheritance):
  "shark-rattle" is a "fish-lure"
  "fish-gorge" is a "fishing equipment";
  "throwing-weapon" is a "hunting equipment;weapon"
- match the string to find its $parent_id. Unfortunately parent terms don't always come before the child term, so 2 passes are needed
- semi-colon separated. Maps to multiple triples
  skos:broader id:thesauri/$parent_id ;
calc_current_hcode: hierarchical codes
- Eg "throwing-weapon" has two hcodes: 2.1.4.F;2.5.D.RE.
- All its sub-items also have two hcodes with matching last component (here *):
  2.1.4.F.*;2.5.D.RE.*
- Uses special chars unknown to me: :, <, >, newline
scope_note: some are a description, others a random note
- uses special chars: \r \n