Re: MiniSKOS update

+1 for ConceptCode.

Regarding the discomfort mentioned, let me offer another perspective. I
have worked on a number of ontologies (at varying levels of simplicity,
depth, and breadth) in which natural language expressions (including names)
are treated as special cases of code: codes in a
code system that is a natural language.

This enabled reusing/specializing infrastructure to handle both artificial
code systems like the examples given and lexical markup. In both cases,
knowing the code system (language, dialect, localization or something like
NAICS code of ICD-9 or LCSH heading for subject) tells you how to interpret
the code (name, alphanumeric identifier string, subject heading string,
linguistic expression, etc.). And that's enough that consumers of the
information can, if they want to, utilize the code/codesystem (including
name/language) information at whatever level of sophistication they want.
That could potentially mean reuse in labeling, matching, using for
lookup/linking, using NLP tools, using "unpackers" for codesystems in which
codes are composed to convey content, whatever.

I am not a linguist, but unless I am forgetting some, each of the systems
(that I've worked on that represent natural language as a kind of code
system and NL names and expressions as special kinds of codes) had
substantial development input by (computational and other) linguists, as
well as ontologists/knowledge engineers, information science folk, and
programmers. I believe that the linguists were the original sources of this
handling, and that it is not uncommon for linguists to think of languages
as code systems (as, for example, in the phenomena broadly known as
"code-switching" (http://en.wikipedia.org/wiki/Code-switching).

To most of the non-linguists among us, this treatment made great sense as
soon as presented, and also made thinking about things like names and
labels and lexicalization less mysterious or intimidating. Hey, we were
already pretty used to thinking about codes and representation of concepts
in code, right? On that basis, I'd suggest that ConceptCode may require
some clarifying text for those not used to folding in NLs under
CodeSystems, but should be among the easier approaches to grasp for most
target users.

Best,
Amanda Vizedom





On Wed, Nov 20, 2013 at 2:48 PM, Dan Brickley <danbri@google.com> wrote:

> On 20 November 2013 19:20, St├ęphane Corlosquet <scorlosquet@gmail.com>
> wrote:
> > I'm having trouble to understand how things like "Wimbledon Tennis
> > Tournament", "Roger Federer", "Tennis" are ConceptCodes, in particular,
> not
> > sure where "code" comes from here, or help. This won't make as much
> sense as Topic for webmasters.
>
> On the other hand, looking at http://schema.org/JobPosting 's
> http://schema.org/occupationalCategory which cites
> http://www.onetcenter.org/taxonomy.html whose values look like this:
>
> 11-9013.00 Farmers, Ranchers, and Other Agricultural Managers
> Plan, direct, or coordinate the management or operation of farms,
> ranches, greenhouses, aquacultural operations, nurseries, timber
> tracts, or other agricultural establishments. May hire, train, and
> supervise farm workers or contract for services to carry out the
> day-to-day activities of the managed operation. May engage in or
> supervise planting, cultivating, harvesting, and financial and
> marketing activities.
>
> 11-9013.01 Nursery and Greenhouse Managers
> Plan, organize, direct, control, and coordinate activities of workers
> engaged in propagating, cultivating, and harvesting horticultural
> specialties, such as trees, shrubs, flowers, mushrooms, and other
> plants.
>
> 11-9013.02 Farm and Ranch Managers
> Plan, direct, or coordinate the management or operation of farms,
> ranches, greenhouses, aquacultural operations, nurseries, timber
> tracts, or other agricultural establishments. May hire, train, or
> supervise farm workers or contract for services to carry out the
> day-to-day activities of the managed operation. May engage in or
> supervise planting, cultivating, harvesting, financial, or marketing
> activities.
>
> These are much more like controlled (enumerated) codes for areas of
> human activity. I can live with "Topic" since you could imagine a book
> being catalogued as being 'about' Farm & Ranch Managers, ... but
> that's not the general intended use of this coding scheme nor it's
> actual use in a job posting.
>
> The 'code' aspect comes from the fact that these are typically
> explicit enumerated lists managed as part of a system. It's not "Farm
> and Ranch Managers" in general, but the onetcentre's taxonomy's
> explicitly coded '11-9013.02: Farm and Ranch Managers' notion.
>
> That said, Peter Mika just raised a similar concern, suggesting that
> e.g. "Arts & Entertainment" isn't really a code. My counter-view is
> that ""Arts & Entertainment"" in some specific news taxonomy
> identified by a standard URI *is* reasonably thought of as a code.
>
> By this point, it's clear that we won't find a name that everyone is
> comfortable with.
>
> > What is the difference between the 'name' and the 'codeValue' of a
> ConceptCode. Maybe some examples would help?
>
> That is a reasonable question. It would also have been a reasonable
> question to ask about the http://schema.org/codeValue of a
> http://schema.org/MedicalCode, but that was hidden away in the medical
> vocabulary where people didn't notice. In many cases it might be the
> same. In some, e.g. numeric subject vocabularies like UDC and DDC, you
> could have a human-oriented 'name' and a numeric 'codeValue'. There is
> also http://schema.org/alternateName to play with now, for alternate
> strings.
>
> For example in UDC, http://udcdata.info/064347
>
> http://udcdata.info/064347
> Notation: 693
> Caption: Masonry and related building crafts
> Including: Plasterer's trade. Finishing work. Tiling. Paving. Asphalt
> work. Composite constructions
> See also: 666.9Gypsum, lime and cement industries. Hard-setting
> materials. Plasters and compositions. Mortar and concrete
> Broader class:  69  Building (construction) trade. Building materials.
> Building practice and procedure
>
> The '693' could be the codeValue, and the caption "Masonry and related
> building craft" it's name. In UDC's SKOS these are skos:notation and
> skos:prefLabel currently.
>
> Looking at the earlier example, I'd suggest similarly
> codeValue="11-9013.02",  name="Farm and Ranch Managers".
>
> Dan
>
>

Received on Wednesday, 20 November 2013 20:29:27 UTC