RE: DDTF/layering: weak class theory seems good enough (5.3, 5.10)

> When Jeremy first proposed[15Mar] avoiding layering
> paradoxes by way of a theory of classes
> in which "nothing necessarily exists,"
>

While the [15Mar] is still my preferred solution because I value a
monotonic layering on top of RDFS; I remind Dan that at the f2f in
amsterdam a straw poll beauty contest was lost 20-3. & I explicitly
withdrew the proposal.

I take it that Dan's message puts my proposal back on the table; and I
will support.

However it is a big uphill struggle to convince enough of the twenty
that the layerability of [15Mar] is enough of a gain to compensate for
the loss of desirable (obvious?) entailments.

Many of the twenty appear to be convinced that:
- layering is a lost cause,
- the layer cake diagram of Tim BL is a vision that we should not even
attempt to take as constraining (i.e. there is no requirement for a
clean, monotonic layering of ontology over RDFS)
- layerings on top of RDF that do not respect RDF Model Theory are in
anyway desirable (even if OWL didn't need it)

I think there are two proposals that I think we can be confident would
work: one in which all the ontology stuff is dark, and one in which no
sets necessarily exist.

I am worried that the position of the twenty is actually backing a
research option - i.e. neither of those extremes is desirable. Pat seems
to say only the lists need to be dark, Peter only seems to need the
restrictions to be dark ... but I don't believe there have been
proposals that have sufficient content to articulate clearly either of
these.

In OWL1 we can choose to:
A: not get all the set theoretic entailments we would like, but get the
layering right
B: get the set theoretic entailments we would like, but screw the
layering
C: take enough time to make a better job on the layering and get all the
set theoretic entailments
D: take even longer and get both the layering and the set theoretic
entailments right.

I think both the last two (C and D) can be postponed until OWL2 (after
this working group).

I personally prefer prioritizing the layering (A), since set theory is
easier.

Jeremy

Received on Wednesday, 29 May 2002 08:07:10 UTC