- From: Sarven Capadisli <info@csarven.ca>
- Date: Fri, 15 Aug 2014 11:12:00 +0200
- To: Gannon Dick <gannon_dick@yahoo.com>, public-lod@w3.org
- CC: KevinFord <kefo@loc.gov>, "public-locadd@w3.org" <public-locadd@w3.org>, "public-egov-ig@w3.org" <public-egov-ig@w3.org>, "public-opengov@w3.org" <public-opengov@w3.org>
- Message-ID: <53EDCEE0.2070609@csarven.ca>
On 2014-08-11 22:58, Gannon Dick wrote: > Sorry for the x-post Don't be. It is a natural thing. > Hi Sarven, > > I noticed you used GeoNames for the "Australian Bureau of Statistics Linked Data" hack mentioned below. GeoNames does much useful work ... but everyone in the Linked Data business could use a little help. > > Domains - in theory, the countries of the world are a group of (federalized data set of ...) (groups of) Court Houses, Jurisdictions, keyed with two and three letter acronyms (ISO 3166). This set for all practical purposes is a Unicode Code Page, but instead of (16x4)=256 members there are (169x4)=676 Latin Alphabet Capital Letters. Statistical metrics at the domain level are manipulated with Linear Algebra and Linear Programming. Diacritics (Côte d'Ivoire) or alternate forms (Ivory Coast) do nothing semantically useful, the acronym is the leveler. > > So, I rewrote the GeoName table (http://www.geonames.org/countries/) to be: > 1) Unicode compliant for XML (HTML entities are HEX escaped) > 2) The Geo's, Country Profiles, whatever are local links. I left those as is and included/matched the MARC System / US Library of Congress Linked Data Service URI's (http://id.loc.gov/vocabulary/countries.html). > 3) Finally, I used an SQL RDB to do an Outer Join on the Code Set - all 676 possibilities. Adding a three character code "synonym" does not increase the code page size. It is then possible to split this "registry" into lists of codes 1) Present, 2) Missing and 3) "Slack" (in the Linear Programming usage). > 4) Put the files in (FODS - (Flat XML) Open Document Spreadsheets format) so that European Civil Servants can not whine about data quality (got your back, DERI, you too ABS). > > http://www.rustprivacy.org/2014/balance/gts/geonames_domains.zip > > Unfortunately, when RDF Lists of Place Names are filtered through previously written applications the result is often unhelpful additions, however these steps should ameliorate the problem significantly. > > --Gannon Thanks Gannon. If I understand correctly, you got around to implement your suggestion back in 2012Q1: http://lists.w3.org/Archives/Public/public-lod/2012Mar/0108.html Care to clarify what I should make of: http://www.rustprivacy.org/2012/urn-lex/artificial-bureaucracy.html ? -Sarven http://csarven.ca/#i
Attachments
- application/pkcs7-signature attachment: S/MIME Cryptographic Signature
Received on Friday, 15 August 2014 09:12:38 UTC