From: Jeff Heflin (heflin@cs.umd.edu)
Date: 03/05/01
First, I apologize for missing the last two telecons: I've been going on
job interviews. Nevertheless, it looks like some good progress has been
made without me... maybe I should miss more telecons! ;-)
Here's my comments on the various specs available at
http://www.daml.org/2001/03/daml+oil-index
daml+oil.daml:
--------------
1) rdfs:range of maxCardinality is
http://www.w3.org/2000/01/rdf-schema#nonNegativeInteger, I assume this
should be http://www.w3.org/2000/10/XMLSchema#nonNegativeInteger
2) ditto for rdfs:range of cardinality, minCardinalityQ,
maxCardinalityQ, and cardinalityQ
3) Why is UnambiguousProperty a subclass of AbstractProperty? Can't
there be Datatype properties that are unambiguous? For example,
numberOfMonth (with domain Month, and range nonNegativeInteger)?
daml+oil-walkthru:
------------------
1) In the section "Defining individuals", it says "Note that to use
datatype values, we had to provide an XML Schema datatype along with the
value. This
datatype is used to parse the lexical representation into an actual
value." I thought that this was optional, especially if the DAML
ontology defines the range. I believe Mike Dean had argued earlier that
if we had to tag the type of each value, then it would be very
inconvenient for users (now come the counter-arguments about how most
users will use tools to do the markup anyway...). Personally, it seems a
little overboard to me, and I'd like to have a choice as to whether or
not I provide the datatype with every value I specify or take my chances
that any consuming agent will download the appropriate ontology before
parsing. If I'm preaching to the converted, then maybe we need to be a
little clearer in the Walkthru document. If this was voted on at a
telecon that I missed, then I guess I can't complain.
reference.html:
---------------
1) In the section "Property restriction", the first paragraph has an
unifinished sentence "Both kinds of restrictions..."
2) My comment for datatype values above (daml+oil-walkthu #1) also
applies to the Datatype Values section
3) My comment on UnambiguousProperty above (daml+oil.daml #3) also
applies in the "Property element" section
model-theoretic-semantics.html:
-------------------------------
1) The text switches into teletype mode during the sentence "The IR
mapping maps abstract properties into subsets of AD x AD and datatype
properties into subsets of AD x DD." and stays that way for the rest of
the document. I assume a missing </tt> tag is the culprit.
2) The document says "A method for asserting the equality and inequality
of individuals would be helpful." Why can't equivalentTo be used to
assert the equality of two individuals?
3) To return to an old debate, is it really necessary to express the
semantics for AbstractProperties and DatatypeProeperties separately? For
example, the document has:
Syntactic Structure: <rdf:type,?R,Restriction> <onProperty,?R,?P>
<rdf:type,?P,AbstractProperty> <toClass,?R,?C>
Semantic Constraint: x in IC(?R) iff IR(?P)({x}) <= IC(?C)
and:
Syntactic Structure: <rdf:type,?R,Restriction> <onProperty,?R,?P>
<rdf:type,?P,DatatypeProperty> <toClass,?R,?C>
Semantic Constraint: x in IC(?R) iff IR(?P)({x}) <= IC(?C)
Note that the only different in the two is that the type for one is
AbstractProperty and the type for the other is DatatypeProperty. Their
semantic constraints are equivalent, so why not collapse them? We know
Property is superclass of AbstractProperty and DatabaseProperty, so we
could state this more simply as:
Syntactic Structure: <rdf:type,?R,Restriction> <onProperty,?R,?P>
<rdf:type,?P,Property> <toClass,?R,?C>
Semantic Constraint: x in IC(?R) iff IR(?P)({x}) <= IC(?C)
Jeff
This archive was generated by hypermail 2.1.4 : 04/02/02 EST