 |
 |
Quoting Ross Singer <[log in to unmask]>: > > Yeah, this could get ugly pretty fast. It's a bit unclear to me what > the distinction is between identical terms in both the geographic > areas and the country codes > (http://id.loc.gov/vocabulary/geographicAreas/e-uk-en & > http://id.loc.gov/vocabulary/countries/enk). Well, in LC's current > representation, there *is* no distinction, they're both just > skos:Concepts that (by virtue of skos:exactMatch) effectively > interchangeable. The distinction is MARC-based. There is a lot of redundant data in MARC that is an encoded form of something that elsewhere is expressed as text -- somewhat controlled text, but text. The geographic area code is input in the "coded data" area of MARC (0XX) to make up for the fact that figuring out a geographic area from LC subject headings is difficult. This is not unlike having publication dates as text in the 260 $c and again in a fixed format in the 008 field. Much of this redundant input (think of the time!) could be eliminated if we quit keying text strings but allowed the display to derive from the coded data. The existence of all of the coded data fields in MARC is proof that there is some consciousness that text is not sufficient for some of the functionality that we would like to have in our systems. Unfortunately, because the coded data is not human-friendly AND is redundant, it does not get input consistently. And because it does not get input consistently, it's hard to base any functionality on it since that functionality would apply only to a somewhat random subset of the records in the database. So... here we are. kc > > See also http://id.loc.gov/vocabulary/geographicAreas/fa and > http://id.loc.gov/authorities/sh85009230#concept. You have a single > institution minting multiple URIs for what is effectively the same > thing (albeit in different vocabularies), although, ironically, > nothing points at any actual real world objects. > > VIAF doesn't do much better in this particular case (there are lots of > examples where it does, mind you): http://viaf.org/viaf/142995804 > (see: http://viaf.org/viaf/142995804/rdf.xml). We have all of these > triangulations around the concept of "England" or "Atlas mountains", > but we can't actually refer to England or the Atlas mountains. > > Also, I am not somehow above this problem, either. With the linked > MARC codes lists (http://purl.org/NET/marccodes/), I had to make a > similar decision, I just chose to go the opposite route: define them > as things, rather than concepts > (http://purl.org/NET/marccodes/gacs/fa#location, > http://purl.org/NET/marccodes/gacs/e-uk-en#location, > http://purl.org/NET/marccodes/countries/enk#location, etc.), which > presents its own set of problems > (http://purl.org/NET/marccodes/gacs/h#location is not a SpatialThing > no matter how liberal your definition). > > At some point, it's worth addressing what these things actually *are* > and if, indeed, they are effectively the same thing, if it's worth > preserving these redundancies, because I think they'll cause grief in > the future. > > -Ross. > -- Karen Coyle [log in to unmask] http://kcoyle.net ph: 1-510-540-7596 m: 1-510-435-8234 skype: kcoylenet