[Top] [All Lists]

Re: [ontolog-forum] LInked Data meme revisited

To: "'[ontolog-forum] '" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Hans Polzer" <hpolzer@xxxxxxxxxxx>
Date: Tue, 10 Dec 2013 17:56:44 -0500
Message-id: <006d01cef5fb$19bb1310$4d313930$@verizon.net>
John,    (01)

Your points are further underscored by the issue of context in which the
words are uttered or published. Much (most?) dialog takes place in
institutional and domain relationship contexts (such as this forum and in
the workplace context most of us have spent a major portion of our lives
in). While the changes in definition of common words are often minor or even
negligible/inconsequential in most such contexts, a significant portion of
the words have very specific and variant definitions in such contexts. Some
of the recent email dialog in this forum provides many examples of this, and
the business world is filled with domain/trade and company-specific
terminology and word senses. Also, while many aim to publish to a broad (but
nonetheless constrained) audience and strive to use "standard" or common
word senses, much dialog is aimed at specific purposes by dialog
participants based on specific relationships among those participants. And
the relationships among participants are themselves dynamic. So we have
dynamism in the evolution of language and word senses, we have broad
variability in contexts in which the words are used, and we have dynamism in
the relationships of the dialog participants to each other.    (02)

Of course, this is all very frustrating to people who want universal
interoperability and understandability - that "universal business language
translator" mentioned somewhat tongue-in-cheek(ly) in a classic commercial
(I believe it was for IBM, if I remember correctly). In theory, we should
all explicitly enumerate all the context, purpose, effective duration, and
frames of reference parameters (among others) that might pertain to the
definitions we use in some given string of words we utter or publish.
Pragmatically, we do it only rarely and only in specific contexts in which
we are aware of, or are alerted to, the importance of doing so. Even then,
we typically only enumerate what we view as the most important context
parameters (sometimes only one). Indeed, we often react indignantly when
someone else points out that we left out some context condition/caveat -
"well of course, that's what I meant", or "everyone knows that", or "I'm not
trying to boil the ocean", or "nitpicker".     (03)

A key function of "session establishment" actions, such as user logon or
"account" creations is to manage dynamism in context and relationships. This
allows some level of consistency and precision in word definitions and data
element definitions in computer-mediated interactions/transactions within
the scope limits of the session or institutional/domain relationships of the
participants (the dreaded data silos mentioned in past dialogs on this
forum).  Rather than continue to strive to do away with such relationship
management mechanisms via universal ("context-free", "relationship-free")
linked data, it would be better, in my view, to add mechanisms for
representing context and relationship information on a "drill-down" basis.
By this I mean to continue to allow context information to be ignored, if
dialog participants or information seekers choose to do so, but to make such
information available "on demand" if participants sense that there may be
some mismatch in word/data definitions. One way to do this is to provide
"meta" links with any data links that point to context/relationship
parameters that drove the particular reason for the link being provided.
These parameters could include a few common context parameters, such as
institution name and linker's purpose for the link, but also should be
open-ended to allow the kind of dynamism discussed above and in your email.
Meta-links would be optional for pragmatic reasons and used only when the
author/creator of the link senses that there may be followers of the link
who should be aware of link context information, but might not be.  I
realize this is a bit of hand-waving on my part, but I believe many existing
domain-specific communication protocols and data models already incorporate
some of these conceptual elements. Let's promote commonality and
standardization  in definitions in contexts of defined scope and purpose,
but support dynamism, diversity and evolution in definitions with changing
contexts, relationships and scope.    (04)

Hans    (05)

-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F Sowa
Sent: Tuesday, December 10, 2013 4:36 PM
To: '[ontolog-forum] '
Subject: Re: [ontolog-forum] LInked Data meme revisited    (06)

Kingsley and Rich,    (07)

The idea of using precise symbols and terminology in science and in
programming languages is useful -- but only for a very narrow application.
The reason why natural languages are so flexible is that a finite vocabulary
can be adapted to an infinite range of applications.  That implies that it's
impossible (and undesirable) to force words to be used with fixed and frozen
definitions.    (08)

> I don't think it will be feasible in the next decade to find a 
> universal dictionary.    (09)

I would revise that point in the following way:    (010)

    It will *never* be possible or desirable to have a fixed dictionary
    of precisely defined word senses for any natural language.  The
    French organized l'Académie française to stop their language from
    evolving.  The net result is that the French adopt their new words
    from the most rapidly evolving of all languages:  English.    (011)

Following is a copy of a note I sent to Ontolog Forum in October.
I strongly recommend Adam K's article.  The title is taken from a comment by
Sue Atkins, a professional lexicographer who devoted her entire career to
defining words and collaborating with linguists, computational linguists,
and computer scientists.    (012)

Many people wish that precise URIs would solve the ambiguity problem.
They could get much better odds by wishing to win the Powerball lottery.    (013)

John    (014)

-------- Original Message --------
Subject: "I don't believe in word senses."  Sue Atkins
Date: Sat, 12 Oct 2013 11:44:01 -0400
From: John F Sowa <sowa@xxxxxxxxxxx>
To: '[ontolog-forum] ' <ontolog-forum@xxxxxxxxxxxxxxxx>    (015)

The subject line is a quotation by the professional lexicographer Sue
Atkins.  She certainly knows what she's talking about, as her Wikipedia
entry indicates:    (016)

    http://en.wikipedia.org/wiki/B._T._S._Atkins    (017)

Adam Kilgarriff, a computational linguist, used that quotation as the title
of a widely cited paper:    (018)

    http://www.kilgarriff.co.uk/Publications/1997-K-CHum-believe.pdf    (019)

 From the abstract of that paper:    (020)

> Word sense disambiguation assumes word senses. Within the lexicography 
> and linguistics literature, they are known to be very slippery entities.
> The paper looks at problems with existing accounts of `word sense' and 
> describes the various kinds of ways in which a word's meaning can 
> deviate from its core meaning. An analysis is presented in which word 
> senses are abstractions from clusters of corpus citations, in 
> accordance with current lexicographic practice. The corpus citations, 
> not the word senses, are the basic objects in the ontology. The corpus 
> citations will be clustered into senses according to the purposes of 
> whoever or whatever does the clustering. In the absence of such purposes,
word senses do not exist.    (021)

I strongly agree with both Sue A. and Adam K. on those issues.  I won't say
that I completely agree with either or both on everything, but the points
they make are always well informed and well worth considering.
Following are Adam's publications:    (022)

    http://trac.sketchengine.co.uk/wiki/AK/Papers    (023)

Annotations can be useful for many applications.  But in general, they must
always be considered approximations for some specific purpose in the context
for which they were developed.  This fact has been very well known to
translators for centuries.    (024)

John    (025)

PS:  Beryl Atkins adopted the name Sue because her husband couldn't
pronounce 'Beryl'.    (026)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: 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 join:
http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (027)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (028)

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