RE: ISSUE-126 (Revisit Datatypes): A proposal for resolution

Oh well, how many databases are there where you have NaN as a field value? Again, NaN is there just to represent the results of
certain arithmetic operations, so, by the same argument as for ontologies, I can't imagine why someone would put NaN into a database
table. Furthermore, I'm really doubtful that such scripts would correctly handle the NaN case: you'd have to have separate cases "if
NaN", "if +inf", and "if -inf". I've written similar scripts myself, but have never really taken such cases into account.


To summarize, I really can't see why we should worry about NaN, +inf, and -inf in OWL 2. This probably got into XML Schema by virtue
of the fact that they based their representation on IEEE recommendations. We don't need to stick to that, particularly not if this
makes our like unnecessarily difficult.







From: [] On Behalf Of Alan Ruttenberg
Sent: 01 July 2008 19:57
To: Boris Motik
Cc: 'Michael Smith'; 'OWL Working Group WG'
Subject: Re: ISSUE-126 (Revisit Datatypes): A proposal for resolution


You miss the point. It wouldn't be put there intentionally in the sense you have it. Rather, one would write a script that packages
up some experimental results, output from some computation. Some of those values might be NaNs. I write such generating scripts (not
usually on number at the moment) quite often.


On Jul 1, 2008, at 1:59 PM, Boris Motik wrote:



XML Schema has already departed from the IEEE recommendation because they don't have +0 and -0. These constants are used to express
problems arising while evaluating numeric operations; for example, dividing 1 with 0 returns NaN.


How many OWL 2 ontologies containing NaN (that are not a test case) are there?


None, because there are no OWL 2 ontologies yet.


Ask again a year after OWL 2 is a recommendation.



Similarly, why would anyone want to say "the weight of individual i1 is +inf"? I haven't seen a single such ontology. Hence, I'd
just ditch these constants and simplify our task. I'm willing to personally pay 1000 bucks to anyone who runs into practical
problems because of that.







From: [] On Behalf Of Alan Ruttenberg
Sent: 01 July 2008 18:52
To: Boris Motik
Cc: 'Michael Smith'; 'OWL Working Group WG'
Subject: Re: ISSUE-126 (Revisit Datatypes): A proposal for resolution


On Jul 1, 2008, at 1:32 PM, Boris Motik wrote:

If we absolutely need +inf, -inf, and NaN, then I'd say we need to add them to owl:real, and then make all other numeric datatypes
subsets of that datatype.


Yes, I was going to suggest that. Since it's ours to define anyways.

Finally, do we really care about +inf, -inf, and NaN? XML Schema might care, but again, XML Schema is a schema and not an ontology
language. XML Schema does not need to do any reasoning on the datatypes; it only needs to perform straightforward validation. This
is why I suggested to change xsd:float: we probably don't want to reason about the properties of floating point arithmetic. We can
keep the name to make people happy. People will be able to put values into their ontology that can be written in the form of floats
and will be perfectly happy with that; for the most part, they won't be able to detect the difference.


The issue is how easy or hard it is to carry data around in an OWL file. If you have some quantity of data and you put it in OWL and
the 3 NaN that are in it make the system not able to proceed because there is a syntax error, and you have to stop and figure out
some ugly workaround,then this is a problem.


It's a little like annotations. They don't have the usual logical semantics, we don't do reasoning on them, but they are important,


I realize that these complicate the reasoning a bit. For one thing, its not clear to me whether we can still say that xsd:float is a
bounded subset of the reals, because of the INFs.




Received on Tuesday, 1 July 2008 19:03:48 UTC