search

duraspace / pcdm

Portland Common Data Model
http://pcdm.org/models
Apache License 2.0
90 stars 9 forks source link

Can http://pcdm.org/models#Object extend http://www.w3.org/2006/gen/ont#InformationResource #13

Closed jcoyne closed 9 years ago

jcoyne commented 9 years ago

This would tie it in with www.w3.org/ns/auth/acl#accessTo which has a range of http://www.w3.org/2006/gen/ont#InformationResource

acoburn commented 9 years ago

My understanding is that rdfs:range is for use by inference engines, not for validation. That is, you can assert: <agent> acl:accessTo <resource>, where <resource> may not already have any existing connection to InformationResource. With RDF, you can make any assertion you want, regardless of the existing rdf:type of a subject/object.

In a word, I don't think this is necessary.

awoods commented 9 years ago

@acoburn, although technically speaking, there does seem to value in being internally consistent given the move towards WebAC as @jcoyne notes.

azaroth42 commented 9 years ago

Actually both pcdm:Collection and pcdm:Object are NonInformationResources due to their inheritance from ore:Aggregation. They have might one or more InformationResources, which are serializations in different formats, such as json-ld and turtle.

I agree with @acoburn regarding range (and domain)... but InformationResource and NonInformationResource are (one imagines) disjoint classes. I agree that WebACL pattern is valuable, but we should be clear about what our IRs and NIRs are first.

jcoyne commented 9 years ago

@azaroth42 Thanks for looking into that. I didn't even notice the inheritance from ore:Aggregation. It does seem a bit odd that an object is a NonInformationalResource as we are putting a bunch of metadata there.

escowles commented 9 years ago

@azaroth42 I don't follow you from ore:Aggregation to gen:NonInformationResource. I'm looking at http://www.openarchives.org/ore/terms/Aggregation and ore:Aggregation is subclassOf dcmitype:Collection. The current dcmitype:Collection (http://dublincore.org/2012/06/14/dctype.rdf) doesn't list any superclasses. What am I missing?

azaroth42 commented 9 years ago

Nothing other than the timeline ... when we were documenting ORE we didn't have the Non / Information Resource terminology or ontology, or at least were completely unaware of it if it did exist.

However from the ORE data model, the NIR-ness is, I think, clear:

A URI-A MUST be a protocol-based URI. However, an Aggregation is a conceptual construct, and thus it does not have a Representation.

From: http://www.openarchives.org/ore/1.0/datamodel#Aggregation

escowles commented 9 years ago

OK, it looks like I've been conflating Aggregation and ResourceMap. And I'm not really sure how ResourceMap would fit in the LDP and/or Fedora contexts. Are the different RDF serializations of the Aggregation different ResourceMaps? Is it OK that they all have the same URI (assuming content negotiation)?