W3C home > Mailing lists > Public > semantic-web@w3.org > February 2009

Re: live meaning and dead languages

From: Xiaoshu Wang <wangxiao@musc.edu>
Date: Mon, 09 Feb 2009 16:14:25 +0000
Message-ID: <49905661.9040101@musc.edu>
To: Frank Manola <fmanola@acm.org>
CC: Jeremy Carroll <jeremy@topquadrant.com>, "semantic-web@w3.org" <semantic-web@w3.org>

Frank Manola wrote:
> On Feb 6, 2009, at 6:03 PM, Jeremy Carroll wrote:
>> Hi Frank
>>> Philosophizing is OK, but some examples might clarify things (or at
>>> least indicate how I've missed the point!).  I will indulge.
>> The example areas that work for me are from social networking.
>> I think we often think about either scientific or inventory type  
>> applications and knowledge, where there is a certain amount of  
>> pretending that things don't change, when in fact they do.
> I think I get what you're referring to here, but it seems to me that  
> (a) there are different "ranges" of mobility of terms (in this sense),  
> and also different degrees of desirability of said mobility.  For  
> example, in your original post you said
>>>> A term which is too tightly nailed down in its relationship to other
>>>> terms has been dug into an early grave. Having fixed its meaning, as
>>>> our world moves on, the term will become useless.
> Does this really apply to, say, "nitrogen" and, if it does, does it  
> apply to the same extent and in the same way as it may, say, to  
> "marriage"?  I doubt it.

I think the issue that we have to understand is how the meaning is 
"nailed down" to its relations to others. Seemingly, "nitrogen" and 
"marriage" are two unrelated *terms". But they are in fact just two 
symbols that relate to something in reality.  In other words, had 
English evolved to use "xxx" and "yyy" to denote the above two concepts, 
they would serve our purpose no less than "nitrogen" and "marriage".  
They are purely symbols,  by itself they carry no meaning.  It is our 
society that assigns them some meanings.  This is the *extensional 
meanings" of a symbol. And now let "xxx" and "yyy" or "nitrogen" and 
"marriage" be URIs.  Then, every URI has an extensional meaning by 
virtual of it being used in our society.

But such assigned meanings is not precise because what the impression 
that the symbol "nitrogen/marriage" evoked in my brain might be 
different from either Jeremy's or yours.  And such a difference, however 
subtle, may increase the equivocation in a communication.  Hence, we 
need to further restrict its meaning for the purpose of clear 
communication, it creates the "intensional meaning" of the term/URI.

The trick and difficulties in the Semantic Web is to manage these two 
kinds of meanings.  I have discussed this issue and proposed some design 
principle and techniques in a paper "Ontology Design Principle and 
Normalization Techniques in the Web", which was published at DILS08 (A 
copy it can be found at 
http://www.inesc-id.pt/ficheiros/publicacoes/4799.pdf). So, I will not 
further discuss their details here.  
>> e.g. I was just having a conversation about the color 'red'. My  
>> colleague was insisting that 'red' is a particular wavelength of  
>> light, whereas I was saying it was the color of blood, coca cola  
>> bottles and fire extinguishers. I think my examples (certainly the  
>> first) predate the wavelength approach to light.
> I'm not sure this example of "red" really illustrates your point.   
> It's not clear that "red" is something whose meaning changed over time  
> (or space) to include these other meanings.  Rather, "red" is the  
> generic, of which these are specific instances.  If I wanted to be  
> specific, I might describe something as "blood red".  If I wanted to  
> be *really* specific, I might identify what kind of blood I was  
> talking about (wet or dry;  mammal or fish; etc.).  But it seems to me  
> "blood red" is a term too;  how much would we expect its meaning to  
> change?  Possibly-interesting sub-example (I don't know which side of  
> the argument, if any, it illustrates):  When I was in graduate school  
> in Philadelphia (Pennsylvania), the police cars were painted red (the  
> fire trucks, interestingly enough, were not), and were slangily  
> referred to as "red cars" (they aren't red now, as far as I know).   
> Anyway, based on this, presumably I could have referred to "red" as  
> "the color of a police car", but *that* certainly changed, and a  
> correct interpretation of the reference would have to include both  
> spatial and temporal coordinates.
>> With social networking sites, certain database fields are displayed  
>> as having a particular significance, e.g. relationship status in  
>> facebook.
>> That these fields are meaningful in the sense of being part of (at  
>> least some people's) life world is clear
>> [1], [2]. The range of choices appears somewhat limited.
>> The choices about which sort of relationship to describe in which  
>> way, will inevitable interact with both social and technical change  
>> (relevant social changes include new concepts like gay marriage,  
>> relevant technical changes include the range of options presented,  
>> and whether these are user customizable etc).
>> I think these sorts of concepts are more relevant to businesses than  
>> the hard-nosed red is the color between r1 and r2 Hz.
> I don't know about that.  For example, if I'm a paint company, I'm  
> pretty interested in that hard-nosed red definition (and so are my  
> customers).  This leads to the idea that one of the things that's  
> important is the agreement on the meaning of terms that exists between  
> the parties in a communication.  If the meaning changes as a result of  
> mutual agreement, that's one thing.  A unilateral changes is kind of a  
> "breach of contract".
>>>> A term which is too tightly nailed down in its relationship to other
>>>> terms has been dug into an early grave. Having fixed its meaning, as
>>>> our world moves on, the term will become useless.
>>>> The trick, in natural language, is that the meaning of terms is
>>>> somewhat loose, and moves with the times, while still having some
>>>> limits.
>>>> This looseness of definition gives rise to some misunderstandings
>>>> (aka interoperability failures), but not too many, we hope.
>>>> So I wonder, as some people try to describe some part of their world
>>>> with great precision, using the latest and greatest formal
>>>> techniques, just how long that way of describing the world will
>>>> last. Maybe there is a role in such precision in allowing us to be
>>>> clear about differences of opinion --- but it doesn't seem to me to
>>>> be a good foundation for building knowledge.
>>> One thing that needs clarification here, it seems to me, is what is
>>> meant exactly by "term" in the SW (maybe that's an illustration
>>> itself!), particularly when comparing it with natural language.  The
>>> meaning of terms may "move with the times", but when a term has been
>>> used in a specific context, I want to know what the term meant in  
>>> that
>>> context (which may be a temporal context, or some other kind).  Take
>>> the term "torpedo", for example.  At one time, it meant a particular
>>> kind of electric fish.  Later, it became used to also refer to  
>>> various
>>> kinds of explosive devices (e.g., naval mines, as in "damn the
>>> torpedos, full speed ahead", and railroad signals).  Today, you
>>> usually think of a self-propelled naval weapon, but the fishes are
>>> still "torpedos", and the use of the term to refer to a naval mine is
>>> obsolete (but would be relevant in, say, discussing the American  
>>> Civil
>>> War).  So is this acquisition (and de-acquisition) of other meanings
>>> an example of the meaning of a term "moving with the times" as you
>>> describe it?  Or do I really have, so-to-speak, several terms "with
>>> the same name" (torpedo).  If the latter, the distinct "terms" seem  
>>> to
>>> have fairly exact and reasonably unchanging meanings (even though the
>>> "same name" can cause confusion).  I'm sure there's some linguistic
>>> vocabulary to describe this stuff.  The point is, I'm not sure I want
>>> "looseness of definition";  what I want is some kind of flexible
>>> versioning mechanism.
>> Natural language assigns multiple different senses to the same term;  
>> often the senses have some relationship to one another, and help  
>> guide further new uses of the term.
>> But each of these senses also drifts with the times.
> It's not clear they always do (at least to any significant extent),  
> but certainly the "torpedo" example illustrates some senses that did.
>> Versioning presents meaning change as a controlled process of steps  
>> - whereas I think meaning change is a somewhat more amorphous  
>> process with no one in control. I just start using a term slightly  
>> differently, and hope it sticks.
> I don't think "versioning" necessarily means this.  But even if the  
> "versioning" is amorphous, I often want to be able to identify the  
> specific "version" of a term that's meant in a particular usage, and  
> distinguish it from others, and I may want a mechanism that helps me  
> do that.  Imagine that the "versions" are partially named by real  
> numbers.  I don't have to assign a version number to every slight  
> difference;  just to the ones I want to distinguish from others.
As Jeremy raised the issue of OWL2, let me use this as an example.  
Because I have not followed OWL2 too closely, I don't know what is the 
detailed policy about the URI for the old concepts used in RDF and OWL1 
and the new one introduced in OWL2, and what about the future 
development of OWL-N. But my guesses is that it won't be the same as 
what is designed follow.  If it happens to be, I will be very glad.

First, I think all terms, such as rdfs:Resource, or owl:Thing etc. 
should be developed as a o3:Vocabulary. (o3's namespace is 
http://dfdf.inesc-id.pt/ont/o3#).  That is: it is defined in a human 
language without any logical assertions.  This defines the "extensional 
meanings" of these term. Because an o3:Vocabulary is inference free, it 
can be maximally shared and any intensional meanings can be built upon it.

Second, any versions of OWL would be the *intensional meanings" of the 
above terms. Each OWL-version would have a URI, where you define the 
logic assertions of the above terms. 

Third, any other specific domain ontologies may simply to "owl:import" a 
specific version of OWL for their own purpose to indicate what kind of 
meanings they intend to achieve.

The above design would make a term's URI stable (i.e., static) while 
allowing their meaning to evolve (dynamic). For instance, even if the 
axioms of OWL-p may contradict with OWL-q about a term "x", the URI of 
"x" needs not to be changed.  And a user is free to choose OWL-p or 
OWL-q and whether OWL-p or OWL-q wins is not totally dependent on its 
design but by its usefulness to its users.

> --Frank
Received on Monday, 9 February 2009 16:15:14 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:45:10 UTC