Len, further to my own reply to you now at
http://ontolog.cim3.net/forum/ontolog-forum/2008-03/msg00254.html,
here are some small qualifications and extensions, but with a
useful pointer at the end, and maybe a big question for you. (01)
I had expressed my hope
> [...] that Pat Cassidy will further explore his
> long-thread-throttled intuition that the very possibility of
> communication should imply some useful level of ontology
> commonality, syntactic and even apparent conceptual mismatches
> notwithstanding. (02)
(Pat, thanks for your immediate follow-up. My response will be in
a later post.) (03)
> My own view, very non-Platonic or non-Pythagorian, is that the
> very definition of abstraction is surely commonality deemed to
> exist between situations, whether between communicating partners
> or between moments of one person. And after all, intension and
> extension should best closely match! (And it is in the end
> immaterial if the commonality / abstraction is discovered or
> invented, ex post facto or a priori, by induction or deduction
> or even abduction.)
>
> Sure, all those generalities might appear remote and useless,
> but I do not at all see a useful outcome of such reflection as
> your "mission impossible". Quite the contrary, I have already
> started to point out how MACK embodies and builds on such
> generalities, and you will see much more of that in my planned
> future posts to this list. That whole scene is even a
> fundamental feature of The Mainstream and its conception of CK.
>
> So the more this forum can consider your and Pat C's "common
> ground" quest, the better it will understand MACK! (04)
(Do skip the first set of new paragraphs now following if you
don't want that rather patronizing recommendation enlarged upon.) (05)
First I must relate how it long ago began to seem so necessary to
do such enlarging. (06)
Ken Baclawski of this forum will recall the wise and charming Haim
Kilov's "Precise semantics" series of workshops at OOPSLAs and
ECOOPs during the mid to late 1990s (and for which Haim produced
full proceedings). During the only one I was privileged to
attend, at OOPSLA'98, there was at one stage a rather lost lull in
the discussion, ended by Haim wondering rather pensively, "Ah yes,
but what _is_ abstraction, really?" He was not expecting an
answer, but I couldn't help piping up, "But it's very simple!
It's what communicating partners have in common." Somewhat taken
aback, Haim considered that for a moment, then nodded slowly in
apparent assent, while looking at me as if in recognition of a
novel observation. Or else it was out of pity for an ignorant or
facile observer? Ken, maybe you have a highly remarkable eye and
memory and can recall Haim's actual impression? Anyway, Haim then
found another way to continue the discussion, and whatever the
true answer, I have further explored, elaborated and built on my
position. (07)
By now I can say that that equivalence of abstraction and
commonality really is the key to MACK with its finetuning of both,
even though it does need fuller amplification as in my first long
self-quoted paragraph above, or as in my point 4 in
http://ontolog.cim3.net/forum/ontolog-forum/2008-03/msg00244.html. (08)
There the very notion of "an entity", even when it is the most
physical thing, is a concept which (whether we see it in 3D or 4D
terms) is abstract in the sense that it 'comprises' or 'embodies'
or 'reflects' - or "abstracts from"! - potentially-multiple
experiences, to produce a deemed sameness or singular "it the
entity" at different times or from otherwise different points of
view, despite continuous change and otherwise inescapable
difference. (09)
That is the core "simplification of complexity" which the MACK
environment builds on, and "Riding The Mainstream" will foster it
in many powerful ways. (010)
Furthermore, and though with this I run the risk of further
confusing you now, you will also see, in a later instalment of
"MACK basics", an extremely useful generalization of that
abstraction/commonality equivalence: even so specific a thing as
one simple triple-entity fact can be regarded, in a "meta-" way,
as equivalent to a MACK Form with precisely one instance of a
FactInContext. That is yet another manifestation of the ideal
equivalence of extension and intension which I noted in the first
self-quoted paragraph above. It is thereby a key aspect of MACK's
reflectivity on its own "meta-" world which is in its turn the key
to a MACK AOS' self-management capability, and hence of the
"application operating system" functionality of the MACK AOS. (011)
That almost-empirical or -phenomenological interpretation of
'abstract' is part of what underlies my "very non-Platonic"
self-characterization above. (So don't be misled by my central
use of the word "Form" in MACK!) (012)
When I get around more systematically to "pillar 3" of MACK in my
"MACK basics" series in this forum you will see how I situate MACK
Ontologically (or is it or "metaphysically" or
"epistemologically"?) as far as philosophical schools or
viewpoints are concerned. (013)
But, ultimately, I shall claim that "The Mainstream Ontology" (the
phrase I perhaps rashly coined "on the spur of the moment" in my
"point 4" I cited above) at least very closely reflects ordinary
scientific and everyday practice. (And please provisionally
excuse such further apparently rash generalization!) (014)
Finally, my usual insistence: once we have a running MACK AOS,
all the above will surprisingly easily be seen as really very
mainstream and obvious! (015)
Meanwhile, it is also relevant to my whole MACK project to add
another personal anecdote from Haim Kilov's OOPSLA'98 workshop.
That was the first occasion on which I had ever heard of Category
Theory, and I was struck by the almost reverential tones in which
the odd allusion to it was made, and by the fact that none of
those present even once dared enlarge upon that obviously-deemed
opportunity. Follow-up there and later soon had me appreciate why
there was both that reverence and that silence. Those
investigations did however also have me note CT's quite possibly
very fundamental applicability to MACK. (016)
Both CT and Mathematical Logic will surely have great scope in
MACK, e.g. for proving correctness and for building with
better-assured scalability. But first let's get the obvious
immediate requirements satisfied! Those formal approaches are R&D
into the future, but the concrete building is largely a much
simpler design and programming problem now, up to the point at
which it can seed the collaborative meeting of many urgent,
massive and already-recognized needs by legions of developer/users
out there in the open market. We can together make great progress
with a seed or boot AOS while yet foreseeing a far more gradual
phasing-in of the products of the formal R&D. (017)
>[snip]
>
>> In my experience software architecture should be bottom-up and
>> scale-free. I am yet to see how MACK can address most basic
>> software architecture requirements, and then achieve high level
>> of abstraction required by Common Knowledge - this is where you
>> need a method which would allow aggregation of simple
>> predicates into complex concepts. How do you propose to do
>> that?
>
> Wow, that's a wonderful question, especially coming from the
> author of www.ontospace.net or rather O(n^2)Space! My answer
> has two very important and counter-balancing thrusts:
>
> 1. Most of my coming 4th instalment addresses precisely the
> accumulation of simple predicates into rich descriptions and
> applications. As you point out, that is essential. But it is
> also a complicating process, and one where many an IT designer
> has been trapped in failure or immobility.
>
> 2. So that is of course where MACK, having been designed "to
> help people simplify complexity together", has many simplifying
> features, largely capturing and implementing natural techniques
> such as the abstraction mentioned above with its various levels
> of subtyping. But a technique which should particularly please
> you as O(n^2)Space author is already illustrated in my
> just-posted 3rd instalment's Customer-Product-Productname-Word
> join: it does exactly your kind of cutting down of semantic
> distance which results in more direct relevance! We even have a
> colloquial name for it: "scrunching", from the way it reaches
> out into logical space and compresses it into more
> directly-relevant joined dimensions, in effect simplifying by
> abstraction once again. My coming 4th or 5th instalment will go
> into that in more detail. (018)
First two qualifications: (019)
a. As intimated in the 3rd instalment, it is of course the
inverse of the above join that covers the facts which give the
answer sought, linking a word to the customers who have bought the
relevant products. (020)
b. My reference to "semantic distance" was taking some liberties
in generalizing that notion. Normally it refers to two ontologies
with a considerable degree of precise commonality already but with
some crucial differences which may be glossed-over. If they are
glossed-over, the above telescoping or "scrunching" of logical
chains similarly produces a precise rendering of some of the
original data but with an analogous (and not equivalent)
glossing-over in the form of dropping knowledge of the
intermediating entities in the chain. Both forms of data
abstraction are simplifications which hopefully do apply to the
situations in which they are used.. (021)
Finally the promised link: (022)
http://www.selectorweb.com/data_warehouse.html with its "Star
Configuration" and "Snow Flake Configuration" is another case of
the usual phenomenon of the mainstream nature of most of MACK's
features, and the familar "MDDB Cube" is exactly an application of
the MACK "scrunching" described. (So conventional BI or Business
Intelligence is also subsumed into the MACK environment! And yes,
the central component of BI's "etl" or extract-transform-load
processes do indeed exemplify MACK's "transformations".) (023)
((Then more relevant personal history is that the software company
I co-founded in 1978, but sold out of in 1990 to start the Metaset
project, became the exclusive South African distributor for
Cognos, billed in the above web page as the makers of "the world's
best-selling OLAP software", building on MDDB as above. It was
however another product we sold and supported, Omnidex (see
http://www.disc.com/), which was the main model for our Metakey
product which was the setting for the "scrunching" described in my
3rd instalment. (But it was Cognos' main product in the 1980s,
the PowerHouse 4GL, which of the specific products I have learnt
and taught provided the most relevant insights and experience now
embodied in MACK and the MACK AOS.))) (024)
And at last a question and a request to you, Len: (025)
In the above I have surely done a great injustice to your
"O(n^2)Space" concept by giving a rather tenuous example of the
kind of phenomenon I believe you have in mind with it. I was
thinking particularly of this from your home page: "writing
software always involves reduction of quadratic complexity as
close to linear as possible." And so on on that page. (By the
way, does that reliably make me think of those famous "six degrees
of separation" between all the people on Earth?) Anyway, both
phenomena seem to me to involve those logical chains reduced to
more situation-useful forms. Though I am sure you mean a lot more
than that, would you agree that my characterization is at least
partially accurate? (026)
Perhaps, though, especially once I have produced my "4th
instalment" and you have had a chance to study it, you could
consider giving this forum an overview of your site's relevance
from an ontological point of view? I would certainly look forward
to such an enlargement of my own horizons. (027)
But now I must get on with that 4th instalment, which will, I
hope, make all the above much clearer... (028)
Christopher (029)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (030)
|