20111019
- select CRM naming variant
- cidoc-ids.txt (CIDOC RDFS)
crm:E1.CRM_Entity, crm:P1F.is_identified_by, crm:P1B.identifies
By the official CIDOC group. This is what we currently use.
- ecrm-ids.txt (Erlangen OWL)
crm:E1_CRM_Entity, crm:P1_is_identified_by, crm:P1i_identifies
- owl2-ids.txt (BloodyBytes OWL2)
crm:E1_CRM_Entity, crm:P1_is_identified_by
Has NO INVERSE properties!
- cidoc-ids.txt (CIDOC RDFS)
Naming in Turtle and Sparql
- "." and "/" are not valid nameChar in Turtle (and neither . The valid ascii chars are [0-9A-Za-z_-], according to the spec and this validatator
- So we cannot use stuff like:
crm:P1F.is_identified_by (because of the dot)
rkd:frame/material/wood/gold_plated (because of the slashes) - Mariana: Absolute URIs are to be used in the problematic cases quoted above.
- Vlado: prefixes are an essential aid for understanding, so we cannot give them up so easily.
While we could give them up for Turtle (which is just an input format), I don't think we can afford to give them up for Sparql (which means writing kilometric queries that nobody can understand).
@base
Barry, you're a turtle expert, can you help? Are there less-anal Turtle processors we can use?
Barry Norton: You have more flexibility (but only one at a time) with the document base URI, e.g.:
But this means changing it within the document.
Vlado: and IMHO is a sure way to insanity
Longer prefix
The best alternative seems to be to have the whole (reflecting the hierarchy) set of prefixes, and to include the dots in these, e.g.:
Vlado: neat trick! But this creates around 250 prefixes... I'd rather select a CRM naming variant that is Turtle-friendly
Naming of thesaurus values
The original idea was to use value URIs having as prefix:
- the thesaurus scheme (type) URI: for values without parent (skos:broader)
- the parent value: for values with parent
Eg:
invalid: rkd:frame/material/wood;
valid but ugly: <http://rkd.nl/thesaurus/frame/material/wood>
valid and better: @prefix rkd-frame-material: <http://rkd.nl/thesaurus/frame/material/> .
rkd-frame-material:wood;
What to do for hierarchical values (eg wood/gold_plated)?
This makes the prefixes too many and too long:
So it's better to replace the last / with eg --:
Mariana: A solution was found that < > escape the . and the / in the uri, and allows to use namespace shortcuts instead of absolute URIs.
Modeling Questions
- which CRM schema defines the Inverse properties?
- document Subproperty policy
If you use has_note and a sub-property has_name, you gain nothing in specificity
then has_note cannot be used for querying, since it will also return the has_name. Therefore all notes should use a subtype - document Subproperty naming
Best us to use the number from CRM and change the prefix and name:rso:P3.has_comment rdfs:subPropertyOf crm:P3.has_note
Visio tasks
- image_objects_carriers@crmg: P65 to Image instead of P62; P138 or P67 to subject
1 Comment
comments.show.hideOct 20, 2011
barry.norton
You have more flexibility (but only one at a time) with the document base URI, e.g.:
But this means changing it within the document.
The best alternative seems to be to have the whole (reflecting the hierarchy) set of prefixes, and to include the dots in these, e.g.: