Re: SUO: Re: More terminology
Hi,
Soem comments, although I'm afraid I won't be able to follow up on this before
a little while. (I've just downloaded Jon's tutorials and I reserved a room in
a monaster for the next two weeks. I'll finally be able to contemplate these
tokens of eternal truth. I haven't decided yet if I'll do flagellation.)
I agree with the intend, and a common terminology is ultimately something we
need (or need to try again to obtain). IMHO, however, there is a risk in trying
to fix terminology in the first place. Precisely because we'll end up in
terminological fight with our eyes blindfolded as concerns the concepts at
hand. Seems to me we should work try to look for the concept first while
remaining very careful about terminology. But once we agree on the concepts and
a system, we can legifere on whichever terminology could work for us,
espectially if some consensus grew during the work. This probably means that
some pieces of the glossary will be fixed provisionally earlier than other...
But I'm a little afraid that a preliminary and working agreement on terminology
on which we would vote, for instance, would come up too much as a constraint
and be counterproductive in the process of working out the ontology. This is
probably arguable and I just might be too cautious here.
As concerns your suggestion for the term 'relation' and 'class'. There is a
large difference between these two here.
'Class' is much more simpler to handle than 'relation'. Because 'class'
belongs, seems to me, exclusively to set theoretical linguo. The only issue
depends on which set theory you assume and which usage this theory generated
historically.
With respect to 'relation# it seems to me much more tricky. Because when
ontologist speak of relations they do not necessarily have a set-theoretic
interpretation in mind. For the concept which is denoted by this term in KIF,
it seems to me that the best terms (at this stage) is simply 'class of
t-uples'. This way we save the term 'relation' from what is a controversial
use.
'Relation' is also used in a sense such that a relation is some sort of
intensional (formal, material, conceptual) object/entity which may have an
extension but which is distinct from its putative extension. The extension of a
relation in this sense is a class of t-uples.
So, to follow up on the 'we should address the concepts first' motto: What do
you think is behind the term 'relationship' in the KIF definition of relation?
Best
Pierre
>
> Terminology management:
>
> John F. Sowa wrote:
> > Pierre,
> >
> > PG>Are you pissed? I apologize.
> >
> > No. I'm not pissed. I'm just frustrated with the endless
> > rehashing of debates about terminology.
> >
> I agree that we need a standardized terminology list for
> this discussion group. I seem to recall that one of us proposed
> to set up such a list over a year ago, but don't recall
> whether anything concrete was decided.
> Can we use this thread to decide how to create a standard
> terminology for this group? Does anyone have experience setting
> up a WIKI? If not, I will volunteer to maintain a terminology
> page on my site (simple list-nothing fancy) until someone
> else sets up a better hypertext version.
>
> As one starting point, I would suggest a careful definition of
> the terms "class" and "relation" and "predicate" (which
> may require definition of "term" and "sentence").
>
> For "class" and "relation" I would prefer the usage that is
> given for KIF classes. It may not be the same as VNBG classes,
> but I think it is the most common use of the term - no?
> The definitions from the KIF site are attached below. Any
> dissenters? (silly question?) If the majority prefer this
> definition, and there is a minority that prefer a different
> usage, I would suggest creating a different term for the
> alternative usage.
>
> Pat
>
> --
> =============================================
> Patrick Cassidy
>
> MICRA, Inc. || (908) 561-3416
> 735 Belvidere Ave. || (908) 668-5252 (if no answer)
> Plainfield, NJ 07062-2054 || (908) 668-5904 (fax)
>
> internet: cassidy@micra.com
> =============================================
> from:
>
http://www-ksl.stanford.edu/knowledge-sharing/ontologies/html/frame-ontology/CLA
SS.html
>
> CLASS
> Documentation:
> A class can be thought of as a collection of individuals. Formally, a
> class is a unary relation, a set of tuples (lists) of length one. Each
> tuple contains an object which is said to be an instance of the class.
> An individual, or object, is any identifiable entity in the universe
> of discourse (anything that can be denoted by a object constant in
> KIF), including classes themselves.
>
> The notion of CLASS is introduced in addition to the relation
> vocabulary because of the importance of classes and types in knowledge
> representation practice. The notion of class and relation are merged
> to unify relational and object-centered representational conventions.
> Classes serve the role of `sorts' and `types'.
>
> There is no first-order distinction between classes and unary
> relations. One is free to define a second-order predicate that makes
> the distinction. For example, (predicate C) could mean that the unary
> relation C should be thought of more as a property than as a
> collection of individuals over which one might quantify some
> statement. Logically, all such predicates would still be instances of
> the metaclass CLASS.
>
> The fact that an object i is an instance of class C is denoted by the
> sentence (C i). One may also use the equivalent form (INSTANCE-OF i
> C). This is not equivalent to (MEMBER i C).
> An instance of a class is not a set-theoretic member of the class;
> rather, the tuple containing the instance is a element of the set of
> tuples which is a relation.
>
> The definition of a class is a predicate over a single free variable,
> such that the predicate holds for instances of the class. In other
> words, classes are defined intentionally. Two separately-defined
> classes may have the same extension (in this case they are = to each
> other). It is possible to define a class by enumerating its instances,
> using KIF's set operations. For example, (define-class primary-color
> (?color)
> (member ?color (set red green blue)))
> Subclass-Of: Relation
>
> ========================
> RELATION
>
> Documentation:
> A relation is a set of tuples that represents a relationship among
> objects in the universe of discourse. Each tuple is a finite, ordered
> sequence (i.e., list) of objects. A relation is also an object itself,
> namely, the set of tuples. Tuples are also entities in the universe of
> discourse, and can be represented as individual objects, but they are
> not equal to their symbol-level representation as lists.
>
> By convention, relations are defined intensionally by specifying
> constraints that must hold among objects in each tuple. That is, a
> relation is defined by a predicate which holds for sequences of
> arguments that are in the relation.
>
> Relations are denoted by relation constants in KIF. A fact that a
> particular tuple is a member of a relation is denoted by
> (<relation-name> arg_1 arg_2 .. arg_n), where the arg_i are the
> objects in the tuple. In the case of binary relations, the fact can be
> read as `arg_1 is <relation-name> arg_2' or `a <relation-name> of
> arg_1 is arg_2.' The relation constant is a term as well, which
> denotes the set of tuples.
> Subclass-Of: Set
>
>
>
--
Pierre Grenon
IFOMIS Uni Leipzig
Haertelstr. 16-18
04107 Leipzig
http://people.ifomis.uni-leipzig.de/pierre.grenon/
pgrenon@ifomis.uni-leipzig.de
phone: 49(0)351971672
fax: 49(0)3519716179
Spe