RDF schemas and Metaweb types topic

see the rdf schema doc

the core rdf schema constructs map almost exactly into the metaweb.

 rdfs:resource - /type/object
 rdfs:class - /type/type
 rdfs:property - /type/property
 rdfs:type - /type/object/type
 rdfs:label - /type/object/name
 

 rdfs:range - /type/property/expects_type
 rdfs:domain - /type/property/schema

these have no equivalent in mql:
 rdfs:subClassOf
 rdfs:subPropertyOf
while you could express subtyping in the metaweb (and there are hints
to the clients to do just that), the query engine itself doesn't infer type
membership - other than the implicit /type/object, all typing must be
represented explicitly.

rdf containers and mql collections are very different.
 - rdf is low-level and more expressive.
 - metaweb collections are simpler.

 - enforcing rdf container well-formedness is put off to the application.
 - metaweb collection consistency is enforced through mql.

 - rdf containers are bags (duplicate elements allowed)
 - metaweb collections are sets (no duplicates) (?)

 rdf containers are themselves resources
  - requires an extra object as a level of indirection.
  - ordering is handled by having a different predicate for each index
 mql collections are just multiple-valued properties.
  - the collection is not itself an object
  - ordering is handled by decorating the links with additional links
  - representation of ordering is less efficient, but a more natural
    extension of unordered collections.
  - possibly more flexible in representing partial orders?

rdfs collections are just lisp lists (first/rest/nil)
 - no equivalent in the metaweb datamodel, but would be easy to add
 - no mql support for these lists though.

todo:
  compare rdfs reification with mql link reification