[Top] [All Lists]

Re: [ontolog-forum] Scheduling a Discussion [was: CL, CG, IKL and the re

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "opensource@xxxxxxxxxxxxx" <opensource@xxxxxxxxxxxxx>
Date: Sat, 12 Jan 2008 13:46:59 +0100
Message-id: <4788B6C3.8030608@xxxxxxxxxxxxx>
Thanks Pat for the excellent summary of the "context" topic. The reading 
list is most welcome and timely.    (01)

I may provide some additional insight why an (global/industry) 
organization such as UN/CEFACT is interested in something like "Context"    (02)

What has been observed in the exchange of business information is the 
that people seem to prefer to reuse what others have developed in order 
to save time and increase interoperability. Almost all reusers say that 
they have slightly different requirements so they *must* (according to 
their industry groups members) add, remove or even change some 
semantics. UNEDIFACT and TDED that was/is the premier KR system didn't 
really handle this situation as wells as desired so when XML arrived new 
ideas was sought after.
A number of key *reasons* for frequently occurring variations was 
identified; business process, product, geopolitics , industry etc. This 
idea of grouping business rules was labeled as Contexts 
Categories,Context Drivers, Context Providers    (03)

One of the questions"ccts:contexts" tries to answers is; Which 
information do I need so send in my message when my receiver has place 
of business in Sweden and sells the product Volvo cars?    (04)

Some members are pragmatist and look for a solution whereby you can 
"categorize / group" the variations and its many rules, statements, 
others believe there exist a context logic rather than partial context 
ontologies, and one person even has a couple of context related patents.    (05)

Currently IMHO the 8 CCTS Contexts  are relevant explanations why reused 
messages do vary, although the defined context are a little too loosely 
defined to be able to be axiomatized. Today there exists no principles 
of how ccts:context are combined and nether EDIFACT, TDED, nor the newer 
CCTS provide sufficient KR primitives to handle these variations 
*within* their formalism. thus complicating buidling something along the 
lines of context ontologies An example is that neither has a 
Generalization-Specialization primitive which efficiently complicates 
representations of variations of aspects of products.    (06)

It would be great, from a practical point of view, to be able to group 
business rule statements according to a few key aspects. One way to look 
at it is as an exercise of selecting the appropriate vocabularies and 
rule sets for particular business situation rather than an working with 
a context logic.    (07)

/Anders    (08)

Pat Hayes wrote:
>>  > [DN]  Would make an interesting ontolog session sometimes in 2008.
>>>  Maybe Peter can schedule a call.
>> [ppy] Sure, I'd be most happy to. I'd need a more refined statement of
>> WHAT (exactly) it is, that folks would like to discuss, and WHO are
>> the people they would like to hear from (or hear discussing the issue
>> at hand).
>> Are we going for CL, CG, IKL and the relationship between symbols in
>> the logical "universe of discourse" and individuals in the "real
>> world"? Representing Context in general? How CL/IKL deals with
>> context? ...
>> I solicit input from those interested on:
>> (a) A clear statement of the "topic",
> FWIW, this is not a new topic. The general idea of the relevance of 
> context in language and KR and logic has been discussed now for 
> several decades, including at least one entire series of successful 
> interdisciplinary research meetings entirely devoted to it. At one of 
> the very first 'ontology summit' meetings (at Heidelburg in, if my 
> memory serves me right, 1997), a whole session was devoted to an 
> attempt to define and clarify the notion. At least four distinct 
> logics for context have been fully developed and at least two of them 
> are now deployed in industrial-scale working software systems. So we 
> are not, or at least SHOULD not be, working in an entirely new area 
> here.
> Unless all the participants are at least marginally aware of all this 
> preceding effort and the general shape of the debates that have gone 
> on, we will only be repeating history.
> FWIW, here is the definition of "context" that I put forward at 
> Heidelberg, and which seemed to meet with some general acceptance.
> -------
> Theories of meaning invariably focus their attention on some subset 
> of the many factors which can possibly influence the actual meaning 
> conveyed by a meaningful utterance or sentence or diagram, or any 
> symbolic structure or act. The other factors or aspects which may 
> influence meaning, but which are not explicitly examined by the 
> theory, are often referred to as "context". Thus, the term "context" 
> has no single meaning: it is used simply to refer to anything (or 
> sometimes everything) which can influence meaning but which is 
> outside the scope of the current theory. It is always defined 
> negatively.
> -------
> This means, in turn, that there is no such 'thing' as an objective 
> context, and so there is no general theory of contexts, or science of 
> contexts, which can be applied to all cases of "context". Indeed, 
> since, in the limit, it seems that almost anything can influence 
> meaning under some circumstances, it follows that anything at all can 
> be considered to be a context. One sees evidence for this conclusion 
> by examining the proceedings of any of the 'context workshops', and 
> checking carefully what each author means by the word "context". At 
> the second such workshop (held at MIT in 2003), every single 
> contributing author had a different notion, ranging in scope from a 
> single occurrence of a single word in a single sentence (as exhibited 
> in so-called 'priming' experiments) to an entire social/cultural 
> milieu, such as China in the 13th century.
> Which, to get back to the topic, is why starting a discussion of 
> contexts with any attempt to "define the topic" is unlikely to 
> succeed, for the very good reason that there really is no such topic. 
> (And even if by some miracle it does, that success will only be 
> temporary, because you can guarantee that someone else out there will 
> disagree.)
> BTW, partly as a reaction to this observation, the formalized 
> 'context logics' developed by McCarthy, Guha (and John Sowa, though 
> he used a different terminology) are deliberately agnostic as to the 
> nature of "contexts", and adopt a theoretical stance along the lines 
> of: a context is whatever satisfies the axioms of my logic/theory of 
> contexts (analogous to a mathematical reply to a question such as 
> 'what is a commutative algebra?'). However, the apparent 
> reasonableness of this 'mathematical' stance is somewhat undermined 
> by the fact that none of the extant such logics provide more than one 
> or two axioms or principles, and the very few that they have provided 
> have obvious, immediate counterexamples in cases that everyone agrees 
> should be possible to treat as contexts (such as a time-interval). 
> This situation has not changed in about a decade of intensive work on 
> these logics, which to my mind at least reduces the 'mathematical 
> theory' position to absurdity: apparently, there is no mathematical 
> theory of contexts. And calling a diverse collection of things by a 
> single name is not good science, nor good engineering, nor good 
> methodology. For example, a good deal of the work on "context logic" 
> has been a slow rediscovery of ideas that were already well known and 
> thoroughly investigated in temporal logic, but because it was seen as 
> being about "contexts" instead of mere time-intervals, the earlier 
> work (which in many cases was vastly superior, having been done by 
> some of the best philosophical logicians of the last century) was 
> ignored.
> The more popular recent stance is that there are many kinds of 
> context, each with its own axioms and principles: time-intervals, 
> time-points, sets of beliefs, linguistic conventions, conversational 
> settings, etc.. Well, OK: but all of these (and others) already had 
> more or less mature fields investigating them, without using the 
> "context" word. So now the question arises, what value is gained by 
> using a single term to refer to them all? It suggests a commonality 
> which does not appear to be present. Time-intervals and states of 
> belief, in particular - the two most commonly suggested "kinds" of 
> context - seem to have almost nothing in common at all. If anyone can 
> suggest a single axiom which relates propositions or sentences to 
> contexts, and is true both when this means the proposition is true at 
> or in that time, and also when it means the proposition is entailed 
> by a set of beliefs, I would be delighted to be shown it. I have made 
> this challenge repeatedly to proponents of 'contexts' for many years, 
> and have yet to be given an answer.
> Pat
>       (09)

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    (010)

<Prev in Thread] Current Thread [Next in Thread>