Re: "In Defense of Ambiguity"

Martin Hepp wrote:
> >Really?  In my experience, "communicating successfully" is a 
> surprisingly rare thing, especially when we are talking about 
> >communicating with precision.
> Well, I for my part feel that humans are able to communicate pretty 
> well if you judge their ability to coordinate joint actions. Most 
> people are able to agree upon a meeting place and time and actually 
> find each other, and when asking a team assistant to come up with a 
> reasonable traveling schedule from A to B, combining any comfortable 
> set of train, plane, car or taxi, he or she will mostly be able to 
> understand and act accordingly.
>
> My computer is not able to understand me at that level.
>
> I would go even further and say that ambiguity in human communication 
> is the oil that makes joint action possible. Human language allows us 
> to be only as specific as is necessary for a given task. Current 
> ontology infrastructure requires that we reach consensus first. Human 
> communication on the contrary allows us to postpone dispute and 
> clarification to a later point in time in which the disagreement 
> becomes relevant, if it ever gets relevant.
This is related to the co-reference problem mentioned earlier. For 
example, what I think co-refers to something else may not be what you 
think co-refers to something else. For example, what I think of as a 
"monument" would to some people include castles, but to other people may 
not include such large physical structures. Yet, this is for most 
purposes not a useful distinction to make. So, mynamespace:monument may 
not be owl:sameAs yournamespace:monument, but we may both classify the 
Eiffel Tower as a monument, but only I may classify Edinburgh Castle as 
a monument. Still, when I tell my friend to meet me at the large 
monument in the middle of town, we both get there usually, since it's a 
statue of David Hume that we all agree is a monument.

All co-reference operates at a level of abstraction. Humans tend to use 
different levels of abstraction for different tasks. Communication can 
co-ordinate action even if people or machines use different levels of 
abstraction. One issue with the (ab)use of  owl:sameAs, as it requires 
two individuals to be logically the same, is that it can merge two 
things that have different levels of abstraction together.

The problem of ambiguity is discussed in depth in IJSWIS 4(2), in 
particular, in the "In Defense of Ambiguity" paper [1]. Note that it is 
not that pessimistic, it merely observes some things about language in 
general, that ambiguity is everywhere in language, but we co-ordinate 
action nonetheless.  Since URIs are like words, the trick may be in 
finding the "right" level of description that allows co-ordination of 
action amongst agents while ignoring - for at least their purposes - 
irrelevant descriptions.

[1]http://www.ibiblio.org/hhalpin/homepage/publications/indefenseofambiguity.html

>
> Maybe we have to discuss "the timing of consensus on the Semantic 
> Web". I think there was a paper by Mareike Schoop and others at the 
> MKWI 2004 conference on respective infrastructure, but I don't find it 
> quickly.
>
>
> Martin
>
>
> Ian Emmons wrote:
>>
>>> However, Hayes is still overly pessimistic about the
>>> difficulties of communicating in the real world.  We humans
>>> have been communicating successfully for thousands of
>>> years.  It is not that difficult to shift our communication medium
>>> to the World Wide Web.
>>
>> Really?  In my experience, "communicating successfully" is a 
>> surprisingly rare thing, especially when we are talking about 
>> communicating with precision.
>>
>>
>

Received on Thursday, 10 July 2008 20:02:52 UTC