RE: SUO: Unanimous consent
Dear Pierre,
See comments below.
Matthew West
Principal Consultant
Shell Information Technology International Limited
Shell Centre, London SE1 7NA, United Kingdom
Tel: +44 20 7934 4490 Other Tel: +44 7796 336538
Email: matthew.west@shell.com
Internet: http://www.shell.com
> -----Original Message-----
> From: Pierre Grenon [mailto:pierre.grenon@ifomis.uni-leipzig.de]
> Sent: 28 August 2003 18:44
> To: West, Matthew R SITI-ITPSIE
> Cc: standard-upper-ontology@ieee.org
> Subject: RE: SUO: Unanimous consent
>
>
> > > But your suggestion has been moved to a motion and the
> > > material to be discussed
> > > as a potential starter on a par with SUMO and OpenCyc, in my
> > > understanding. And
> > > for this purpose, as you hinted yourself, the material is
> > > immature.
> >
> > MW: As far as maturity goes, this work is part of a stream
> > that goes back at least 15 years - which puts it on a par
> > with CYC, and somewhat ahead of SUMO. However, it is not as
> > completely axiomatised - which is a different question.
>
> PG: I was refering to the second kind of maturity. Not the
> maturity of the
> project's findings (anyway, longevity is not always a sign of
> maturity, you
> will agree), but that of the document in its capacity as a
> proposal for being a
> starter.
MW: This is a clear difference of opinion that we have. You
presume that a starter document must have a certain level of
completeness. I do not. No rules have been set by this group
for the level of completeness a starter document should have,
so you can only be extrapolating from SUMO (in fact IFF was
substantial, but rather incomplete when it was first propsoed).
I think your presumtions are questionable as "rules" of this
group, although you are of course entitled to your own
preferences as to the level a document should be. In the end
the group will decide.
>
> > MW: It is not clear to me that any particular degree of
> > completeness is required for a starter document. Pretty
> > much by definition a starter document is not finished.
>
> PG: Agreed, of course
>
> > In ISO
> > where I am used to the way things work, you can make a New
> > Work Item Proposal with either a statement of the intended
> > work area, or with anything up to a finished document. The
> > only thing that is important is to be clear about what you
> > have as a start point and where you are going. I think I
> > meet those criteria. As far as I can see the main reason
> > to make a proposal is to try to gather support (and possibly
> > effort) around doing some work.
>
> PG: We could discuss this at length I imagine. Seems to me that a good
> candidate starter would comport a reasonable amount of
> axiomatization (at least
> the more fundamental features of the claims), doesn't mean that the
> axiomatization needs to be perfect either. The work of the
> group would be to
> correct as needed, expand, compare, and so on.
MW: Your opinion again rather than something already agreed by
the group. I differ of course or I would not have made a proposal.
>
> > > I would
> > > have seen no reason to vote on accepting this as an expert
> > > contribution and
> > > certainly no reason to reject this. As a starter, I would
> be much more
> > > cautious.
>
> PG: Not sure if the above was clear. The point is this may be a good
> contribution but seems to be an underspecified starter.
MW: Again you opinion against your criteria, not something
laid down by the group.
>
> > MW: I think there are two key issues to consider here:
>
> PG: These questions are releveant but not directly to whether
> your material
> ought to be a starter or not, imho. My previous comments were
> not discussing
> the actual theory.
>
> > 1. Should there be a 4D ontology as part of the SUO work?
>
> PG: There should be an ability to sustain a 4d ontology.
> I.e., there should be
> an ability to deal with 4D entities. I think there should not
> be any claim
> about which are what.
> Your approach seems to provide a 4D ontology. It is, in my
> opinion, faulty in
> that it commits to 4D. (Not sure what i'm saying is very clear.)
MW: Committing to 4D is the intent. That is not supposed to commit
the whole SUO effort to 4D though.
>
> > 2. Is this a good start point? (if not suggest a better one)
>
> PG: It might be, told you, can't tell. It actually depends on
> the relationships
> there are and how they would be axiomatized.
MW: A good reason for doing this under group control.
>
> > > In my humble opinion, the way to proceed would be to withdraw
> > > the motion and
> > > simply register the material linked in a couple of your
> > > messages (browser and
> > > flat file) as expert contribution. Then, after you have done
> > > the work of
> > > turning this into an axiomatized full fledge ontology
> > > comprable to SUMO or
> > > OpenCyc, it seems to me that this could be an interesting
> > > candidate for a
> > > starter and that a motion could be submitted at that later time.
> >
> > MW: I would rather have an indication of support, and perhaps some
> > help doing it.
> > >
> > > The problem as I see it is that, as it stands, this material
> > > is a potentially
> > > interesting contribution to the initial declared objective of
> > > this group of
> > > coming up with a SUO by contributing to the development of
> > > starter documents.
> >
> > MW: This is all we have at the moment.
> >
> > > (I think it would profite the development of OpenCyc based
> > > document more than
> > > that of SUMO). Incidentally, and to gloss on my own soup, it
> > > seems to me that
> > > this is something which could communicate well with one part
> > > of our framework
> > > here (that which Barry Smith as baptized SPAN as you know),
> > > provided we
> > > recognize that terminology is not an issue (I could be hung
> > > on the marketplace
> > > in Leipzig for saying this publicly).
> >
> > MW: Actually it would be even more interesting to see how to
> > communicate with the other part (SPOT?).
>
> PG:
> I'm sure Barry could find something to name SPOT. It's
> SNAP... for snapshot.
MW: Sorry, I couldn't remember and was too lazy to look it up.
> It's presentist, 3d, good old fashion Aristotelianism. The
> whole thing in which
> SNAP and SPAN (which is primarily a temporal distinction) are
> articulated is
> BFO.
>
> I think SPAN (our 4D) can handle much of your ontology. Your
> ontology relates
> to SNAP the way SPAN and SNAP relates. In our official
> linguo, your physical
> objects are parts of lives of SNAP entities, they are SPAN
> entities themselves.
MW: Something like that, but of course 4D just doesn't bother
with the SNAP stuff - representing information just with a SPAN
type objects.
> (Of course, calling them physical objects is pure
> terminological heresy.)
MW: (Surely the heresy is SNAP :-) )
> It
> seems that you handle participation as parthood, which is
> coherent with my BFO
> parsing of your claims. We don't, there's no parthood
> relations between SNAP
> and SPAN (apologies for the terminological endoctrination).
MW: Correct.
>
> One difficulty is that you have processes with qualities
> (i.e., the t.p. part
> of your car is red). It would be odd for us to think of
> redness as a property
> of your car's life.
MW: They are not necessarily processes in our ontology, so
this objection makes no sense.
> What we mean is that the car's life is
> that of an entity
> (the car) which happens to be red during the occurrence of
> one of the temporal
> parts of such process (the life). The redness of the car (a
> particular) is on
> the side of continuants.
MW: Well to justify SNAP you have to decide when you are
going to use them, but 4D sees the world differently, and
does not accept these "rules".
> Now, you can certainly introduce as many process-types as you
> want, but this
> seems rather ad hoc. This is a general feature of processes
> which have their
> properties necessarily and which let themselves best
> interpret as instances of
> classes (class nominalistic properties). I didn't have time
> to look cloesly at
> the way you handle properties, pardon me.
MW: Properties are classes, individuals are members of them.
>
> > > Re. Axioms. Of course, the taxonomic constraints are
> > > expressed axiomatically,
> > > but this is not a strong claim. It's a minimum requirement.
> > > This is why a DL
> > > would probably be enough to formalize the current state of
> > > the work. To be
> > > fair, there are also a number of typing constraints on
> > > relations, but again,
> > > this is sort of minimal requirement. Again, for this it seems
> > > that the DL
> > > vocabulary of OWL could be enough. The problem is that
> > > taxonomical axioms give
> > > you a taxonomy and not an ontological theory.
> >
> > MW: Then at least we can say that there probably aren't many
> > mistakes yet :-)
>
> Nice try, but it's probably a very sinful situation to commit so many
> omissions...
>
> > >
> > > So, with respect to whether this is a good 4D ontology. I
> > > just can't tell.
> > > There are good clues, such as your all encompassing domain of
> > > possible_individual (which is scary term by the way) and some
> > > hints (mostly
> > > implicit?) of an interesting mereological approach, although
> > > I'm not sure I
> > > really see what temporal_part_whole exactly is. But it is
> > > hard for me to tell
> > > after looking at this less than a handful of times. As far
> > > curiosity goes, it's
> > > not clear how events relates to non event possible_individual
> > > (are they parts?
> > > I can only remember something about referential involvement,
> > > there may be
> > > another relationship). But, at the moment, this is
> > > essentially documentation
> > > exegesis, a major problem for me is the lack of an
> > > axiomatization of these
> > > concepts.
> >
> > MW: If you want to understand the ontology and where the
> > axioimatisation will go I suggest looking at:
> >
> http://www.matthew-west.org.uk/Documents/Spatio-temporal-Paradigm.pdf
>
> I've seen it before, but I had forgotten about it.
>
> > MW: I'm updating it at the moment, but mostly about references
> > rather than basic content. That should help with temporal whole/part
> > (usually used to indicate that some possible individual - a state -
> > is a temporal part of some whole_life_individual).
>
> PG: One of the problem is the way you handle relations
> between physical object.
> It seems that all relations are cases of underlap in a
> process (or activity,
> unless that's too specific).
MW: Perhaps in SPAN terms, but in our terms they are states that are
related to each other.
>
> While we're at it and to go abck to earlier things about
> participation. Seems
> to me that another problem with the paper is in the very
> claim that activities
> are aggregates of individuals, e.g., a meeting is the sum of
> the participants.
> Not sure that all parts of the participants are parts of the
> meeting the way
> the participants are parts of the meeting.
MW: The participants are temporal parts rather than whole life
individuals.
> There's some form
> of shading which
> occurs and can block transitivity on occasion. Maybe this is
> debatable, it also
> has to do with modality I guess.
>
> Yet another related issue is that you collapse identity into
> spatiotemporal
> colocation.
MW: Correct. It seems to be the "norm" with pure 4D ontologies.
> I think we discussed that already a few weeks
> ago. As I said at
> that time, this is a strong rigidity of your approach
> (incidentally, it is not
> a general feature of 4D).
MW: It is not a necessary choice I agree, but it is Ted Sider's
choice, amongst others. The rigidity has advantages of course.
>
> > MW: I believe in any case that you should write things down in plain
> > text before writing the axioms.
>
> The problem is to know what to leave in plain text and what
> to formalize (a
> slightly different problem is what can be left and what can
> be formalized).
MW: I agree.
>
> > MW: The full FDIS document also provides a great deal more material
> > than the model, and I commend it to you as someone seriously
> > interested. It should be published in the next few weeks. You can
> > probably get it through the Univeristy, but I'm afraid ISO it has
> > to be bought.
>
> PG: I don't know what 'FDIS' means. Are there any preliminary
> public versions
> approximating closely enough the final document and logs of
> the review process
> available?
MW: Final Draft International Standard. I'm afraid even the previous
version (Draft International Standard) had to be bought. The logs of
the review process are available, but not much use without the base
document.
>
> > > (Which is also why I'm leery about the capacity of
> > > this document to
> > > serve as a starter, which, again, seems to be the question at
> > > stake with the
> > > motion.) It seems that I have to imagine what kind of axioms
> > > you will write in
> > > order to understand what kind of theory you propound.
> >
> > MW: Not really. By making it a starter document I am actually in
> > principle putting the control of the addition of axioms with the
> > group, rather than being just what I think.
> >
> > MW: It is not particularly how SUMO and IFF have worked in practice,
> > but I would certainly expect to make proposals for axioms
> in particular
> > areas that were under the scrutiny of the group, rather than my own
> > whim. I would be editor (perhaps) rather than author.
>
> PG: As I understand it, this group aims at authoring a
> standard, not authoring
> candidates.
MW: All standards start as candidates. You don't get one without the
other.
> There is a risk of counter productivity here.
> From your standpoint
> I understand this would be productive. From the group's
> standpoint it seems
> that this would mean division of workforce and less focus on existing
> documents.
MW: Hardly. Those that are intereseted in pursuing a pure 4D ontology
really have nothing they would do actively at present. Those that
want to pursue a 3D, or other, ontology are hardly likely to be
diverted. As a person withbroad interests you might be the exception.
>
> P
>
> >
> > MW: I would also expect to keep an issues log on the document.
> > >
> > > Best,
> > > Pierre
> )+
>