[Top] [All Lists]

[ontolog-forum] Five challenges for "semantics" beyond Knowledge Represe

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "Phil Murray" <pcmurray2000@xxxxxxxxx>
Date: Mon, 29 Sep 2008 10:44:46 -0400
Message-id: <a94f2fc20809290744h51964b6fne0453a6f827ef0d4@xxxxxxxxxxxxxx>
The viewpoint of the community represented by the Ontolog forum is
knowledge representation (KR), with an emphasis on enabling computers
to support and make use of such knowledge representations. In
particular, as Christopher Menzel observes, "KR exists precisely
because you can't rely on informal, intuitive understandings of
English when you want to share and process information, and use
computers to aid significantly in the process." And John Sowa places
the emphasis squarely on computing: "...any ontology and reasoning
method that is relevant to engineering, computer science, AI, or the
Semantic Web must be computable." More specifically, we are talking
about information on which "logic-based reasoners" can operate.    (01)

As an outsider, I understand the necessary bias of the KR community.
That's what you are trying to do, after all, and you have, as a
community of experts, done a lot of brilliant stuff for this purpose.    (02)

*My* interest, as a member of the non-profit Center for Semantic
Excellence, is making *work* better in the face of an overwhelming
amount of information. I believe that "semantics" has an important
role to play in knowledge work overall.    (03)

In a recent post, Pat Hayes observed, "Semantics isn't about knowledge
capture, or database management, or the Semantic Web, or about the
future of Western Science, or any of that grand stuff, though it might
well be relevant to those topics." Well, I think making work better
and making knowledge workers more effective *is* grand stuff, and that
semantics is directly relevant. (And, hey, isn't knowledge capture
rather, er, meaningless if there's no meaning involved?)    (04)

Challenge #1: What do we call this?
-----------------------------------    (05)

Or at least insofar as semantics is part of what John Sowa refers to
as "rhetoric" or "pragmatics." That's the first part of the problem.
What terminology should we use for the day-to-day transfer of ideas
within groups or organizations, the communications that include
informal discussions and unstructured documents? Is it "rhetoric," or
does it need a new name and a new definition that reflects the impact
of the Information Age on human communications? And how should
semantics be associated more clearly and more widely with those
activities?    (06)

Examine any document or oral communication -- including the posts to
this forum. It's all about communicating ideas. Yes, we define
terminology on the fly, too, to achieve disambiguation, but once the
meaning of the terminology is agreed upon, we move on to talking about
the connections in our immediate reality (not theories of representing
reality), evaluating those ideas, selecting the important/relevant
ones, connecting those ideas (chains of causality or influence), and
remembering all of those things.    (07)

In the everyday world, people are concerned with Who, What, Where,
Why, etc. A more precise and useful description would be to break down
those ideas into Agents, Actions, Inputs, Outputs, Instrumentality,
Goals, and "Context" (at least in some crude form). Because that's
what matters in most work activities. And in most cases it doesn't
matter whether an idea containing such elements of reality can be
parsed against a formal model of reality (one that supports
inferencing), as long as people can create, find, understand, and
re-use such representations easily and productively.    (08)

This is, after all, what we do all day. And it's a much shorter leap
for most knowledge workers to understand the nature and impact of
"ideas" than it is to grasp theories of representating static
realities. As Bob Glushko observed a few years ago, "We need 'semantic
literacy' or maybe even 'ontological literacy' but maybe we don't
teach it because it is too hard to explain what they mean." We do need
greater understanding of semantics, but we can't convert everyone into
ontologists. And we shouldn't. (And, besides, the term "semantic"
scares the bejeebers out of most people.)    (09)

The first wave of computer literacy focused on the tools. The second
wave focused on information, with desktop publishing and the Web
making us all both beneficiaries and victims of the Information Age.
The third wave must be centered around meaning and how it produces
value.    (010)

Challenge #2: Should we call these communications "ideas" or ...?
------------------------------------------------------------------    (011)

The second challenge I encounter is what to call "ideas" themselves.
There's a host of similar but tainted expressions in English.
"Propositions" or "statements" may be comparable to "ideas."
"Assertions" seem to have a special meaning in KR; they imply or
stress support for inference.    (012)

And of course, it does make a difference whether we're talking about
natural-language sentences or structured representations of the
meaning of those sentences. "Sentences" is also burdened with the
quirks of natural language, it seems to me. But maybe it's on the
right track if Simon Williams chooses to call his product "Sentences."
(See http://www.lazysoft.com/.) The intent seems right, but I'm uneasy
with it. See also "sentences" in IKL
(http://www.ihmc.us/users/phayes/ikl/guide/guide.html)    (013)

Maybe we need to borrow a term from a foreign language, as non-English
speakers do so frequently from English. Turnabout is fair play.    (014)

Challenge #3: Inadequate tools
------------------------------    (015)

The third challenge is the absence of an adequate set of tools for
visualization and management of ideas. That may be key to acceptance
of semantic literacy in general. We are, after all, intensely visual
animals. The value of words in our evolution cannot be understated,
but an overabundance of words coupled with too little time to consume
them tends to overwhelm and hide meaning. It often fails to provide
the kind of "Aha!" moment that graphics do. Visualization tools also
*force* us to be explicit about objects and relationships. UML,
Conceptual Graphs, and Petri nets may be first steps, but we need
combine the best of each and move forward.    (016)

The visualization tools available in such applications as Compendium
and Cmap are helpful when relating and discussing ideas ... but
limiting. Somehow, we've got to get beyond two-dimensional
representations, and we need to combine the fine-grained understanding
of vocabulary made possible with taxonomy/ontology with even greater
-- and easier -- expressiveness of relationships among "ideas."    (017)

Challenge #4: We are an unknown
-------------------------------    (018)

The fourth challenge I see is overcoming the assumption that people
are naked novices every time they dive into the world of information.
For the most part, we assume that people looking for information must
engage in a semaphore-like "conversation" with a search engine ...
every time. It's more like one-dimensional Scrabble than a
conversation. The search engine knows very little about us. And when
it sends us information, we have no pockets to put that stuff in.
(Google Desktop scares me at times with how much it seems to know
about my search terms, but it's still all about funneling my
interaction through that little box, one string at a time.)    (019)

This situation doesn't make sense. When you're arguing with your
co-workers about ideas, they learn (over time) what you know, what
your biases and interests are, and even your preferred mode of
learning. In order to participate effectively as an individual in the
broader "meaning-filled" world, individuals must be equipped with
tools for interacting *efficiently* with that world. Whether we call
this having a "personal profile, a "user model," or a "personal
ontology," the need for such tools is largely ignored in practice. And
although we have extended the notion of personal profiles into
avatars, avatars have to live in a time-based world.    (020)

Having a "user model" seems to me not only a good idea, but a
necessity. How does it help you if the search engine has to guess what
you mean every time you execute a query? Do you reorganize your files,
bookshelves, and writing tools from chaos to your customary
arrangement in order to suit your regular work habits every time you
go to work? (See Judy Kay's "Scrutability for Personalised Interfaces"
for one limited look at work in this area.)    (021)

Challenge #5: Describing what people do ... not what you think they do
----------------------------------------------------------------------    (022)

The fifth challenge is our need to integrate a "what-if" capability
into the representations of what people do. Aw Kong Koy of
Multicentric Technologies -- a friend, engineer, and all-round good
thinker -- appears to have been the first to observe that, "You can't
manage what you don't describe." This is a simple but remarkably
astute twist on that old saying, "You can't manage what you can't
measure."    (023)

Both observations are true. Pretending that you're managing knowledge
workers by looking at their outputs (unstructured documents, in many
cases) at the end of the month isn't really management at all. You
have to know what they're doing, how often they're doing it, the
instruments they use, the outputs of their activities, and the
constraints (for example, regulations) applicable to their work -- at
a granular level. And you have to know when those activities change.
Describing *what* people do is a semantic perspective. Attention to
semantics makes it possible to describe what people do in a structured
way. If you as a manager don't know what your people are actually
doing, then you're just shouting at the river to try to change its
course.    (024)

If you *do* know what people are doing, then you can project how
changes in what people do will affect the outcomes of their work as a
group ... at least to some extent. That's way better than shouting at
water.    (025)

Conclusion (at last)
--------------------    (026)

Phew. Apologies for the length of this post. But I think the KR
community has at least some of the responses to these challenges. I
believe the KR community can benefit people and the economy more
directly and much more broadly than it appears to do at the moment.
Help!    (027)

Phil Murray
Secretary and Senior Research Fellow
The Center for Semantic Excellence
pcmurray2000@xxxxxxxxx    (028)

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    (029)

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