Re: (Round 2) Proposed Extensions to OWL

[Roger L. Costello]
>
> 2. There is a relationship between
>
>         concat(rdf:value, units) in resource #1, and
>         concat(rdf:value, units) in resource #2.
>
>     "There is a relationship between the concatenation of the
>      value of the rdf:value property with the value of the units
>      property in resource #1, and a similar concatenation of
>      values in resource #2."
>
>     The relationship is:
>
>        1.0 inch = 2.54 centimeter
>
> Do you agree that this is the relationship between the two
> anonymous resources?

No, I do not. Trying to stick close to the way you have set this out, there
is not a direct relationship between the __concatenations__ of unit and
value.  You could say there is a relationship between the __tuple_s_  (1.0,
inch) and (2.54, cm), but that is still superficial because it depends on
the number of significant figures and the round-off strategy.   I would
rather have it be more fundamental.

Here is what we know, it seems to me -

1) Both resources are length measures.
2) The value of a length measure can be expressed numerically in different
units.

(This is a scalar quantity.  A more complex quantity, like a vector or
tensor, would have to be expressed as some structured value).

We know a few other things, but they can be formulated in various ways.
Here is one way.

3) The numerical value of a length measure may be obtained by applying an
operator (or a transformation) to it.  That is, conceptually
length-in-inches = L1 * M, where L1 is the operator for getting the length
in inches, and M is the measure.

4) A measure may have any number of such operators or transforms, one for
each different unit of measure.

Here is a set of triples (minus namespaces) for your two resources that
captures, I think, the essence of these points -

{resource
    {type:length-measure}
    {value:
        transform:
            {
                type:length-in-inches}
                number: 1
            }
    }
}

{resource
    {type:length-measure}
    {value:
        transform:
            {
                type:length-in-cm}
                number: 2.54
            }
    }
}

I think that these are admirably simple, and I doubt that you can simplify
them any more without losing their essence.

The resource type can obviously be stated in OWL as part of an ontology, and
so can the transform type.  If we had a standard way to make math statements
with OWL, we could make an OWL statement that the two transforms (inch and
cm) had some kind of "equivalentValueTo" relationship.

This approach takes the matter of the relationship between length in cm and
length in inches away from the individual resources and puts in onto the
expression of the relationship between the transform types.  I think this is
very appropriate.

The upshot is that you need to come up with some convention for expressing
the relationship between the transforms - or of testing for equivalence -
and then everything else can be handled in OWL.

Cheers,

Tom P

Received on Friday, 27 June 2003 09:48:01 UTC