You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
examples are the most important thing, so please include more
All URLs should use the same host name. The examples you give are ok in this respect, but you also write "divide the root domain name" (not ok). In comments on gsbpm-ontology.ttl #10 you use rdf vs id in the host name (not ok)
URL parts should be in singular since each URL describes a single thing. Eg model not models, concept (or thesaurus) not concepts. You do this for later parts (eg class, section): why not for earlier parts?
nacer2 is hard to read. Use a separator or capitalization to show the two parts, eg nace_r2 (preferred by me) or NACEr2.
for correspondence tables, use URLs like /correspondence/isic_r31/nace_r2 or /correspondence/ISICr31-NACEr2
rather than concept for all cases, I'd use more specific URL parts, eg classification/activity vs classification/product. But you should be the judge of that since you know the total scope of data.
each classification should also have a canonic URL always pointing to the latest version. Eg at present /classification/activity/nace should be the same as /classification/activity/nace_r21, but in the future it should redirect to the newest version. Why: because company data in registers rarely specifies the version of a classification, so to represent it most faithfully and stably we should use version-independent URLs
for classification concept URLs, do not use a part designating the level (section, class etc). Why: because devs will appreciate the opportunity to define and use a single prefix (eg eu_nace:55.11 (a class) or un_isic:B (a section)), rather than having to use full URLs or numerous prefixes. These designations are "somewhat random" anyway: you call them "Resource type" but these are all skos:Concepts in a single hierarchy.
for meta URLs, the trailing part should mirror as closely the resource being described, eg: /meta/adms/classification/product/un_cpc_v21
The text was updated successfully, but these errors were encountered:
https://github.com/FranckCo/Stamina/blob/master/doc/content.md:
rdf
vsid
in the host name (not ok)model
notmodels
,concept
(orthesaurus
) notconcepts
. You do this for later parts (egclass
,section
): why not for earlier parts?nacer2
is hard to read. Use a separator or capitalization to show the two parts, egnace_r2
(preferred by me) orNACEr2
./correspondence/isic_r31/nace_r2
or/correspondence/ISICr31-NACEr2
concept
for all cases, I'd use more specific URL parts, egclassification/activity
vsclassification/product
. But you should be the judge of that since you know the total scope of data./classification/activity/nace
should be the same as/classification/activity/nace_r21
, but in the future it should redirect to the newest version. Why: because company data in registers rarely specifies the version of a classification, so to represent it most faithfully and stably we should use version-independent URLsun_isic_r31
,eu_nace_r2
,bg_nkpd_2008
. This is inspired by http://api.opencorporates.com/documentation/API-Reference#industry_codes so cc @openc and @bensymmondsontology
for ontologies: notmodels
nordef
section
,class
etc). Why: because devs will appreciate the opportunity to define and use a single prefix (egeu_nace:55.11
(a class) orun_isic:B
(a section)), rather than having to use full URLs or numerous prefixes. These designations are "somewhat random" anyway: you call them "Resource type" but these are allskos:Concept
s in a single hierarchy./meta/adms/classification/product/un_cpc_v21
The text was updated successfully, but these errors were encountered: