Re: Cross-ontologies reasoning

>Pat, you wrote:
>
>  > But the real reason for optimism is that there
>>  is no reason not to do this, and every reason to do it.  Human
>>  language evolved because it is useful to be understood and to
>>  understand: the ability to communicate benefits both ends of the
>>  communication channel. So writers of ontological content for the
>>  SWeb, and users (readers) of that content, will all feel the economic
>>  pressure to re-use existing content as far as possible, so that they
>>  can be understood and can understand one another. All this pessimism
>>  seems to me to be like worrying that if people were all to invent
>>  their own language, communication would be very difficult. Which is
>>  true, but only relevant if there is any reason to think that people
>>  are likely to do that: and there isn't.
>
>I appreciate your optimism, but I prefer to maintain a healthy 
>skepticism in this area. (Too much optimism, after all, is what 
>caused the disillusionment and backlash of the "AI winter").

That was caused by too much BOASTING, not too much optimism. And we 
got over it, eventually. Still, I agree that scepticism is a useful 
strategy, provided it is informed scepticism :-).

>The subject of human communication is, of course, extremely complex 
>and we could spend endless time on it. Let me just point out that it 
>is not always true that people strive for communication. There are 
>so many examples of language variations intended to communicate only 
>within a restricted circle of people at the exclusion of everybody 
>else, or not to be understood, period (just think of some political 
>jargon).

All true; but (1) ontology sharing isnt anywhere near as hard as 
human communication; and (2) suppose the scenario you suggest happens 
on the SWeb: is that a problem? If subcommunities do not wish or need 
to share meanings, then the fact that they do not is not a problem.

>
>But let's look at something much simpler and down to earth: the goal 
>of code sharing. It has been discussed for so many years within the 
>computer community, and in theory it makes so much sense. But the 
>reality has been very disappointing so far. Ontology sharing has 
>some of the characteristics of code sharing, and I would like to 
>understand what would make ontology sharing much more successful 
>than code sharing.

Good point. My knee-jerk reaction is that ontologies do not need the 
maintenance of any kind of state other than that which is expressed 
in the ontology itself. This is why assertional languages are 
inherently more transparent to context than almost all kinds of code. 
It is quite hard to keep the formalisms 'pure' in this way (witness 
the current debates about using negation-as-failure in SW rule 
languages) and I would agree that a tendency to incorporate assumed 
computational states into the formalisms is a danger which requires 
constant effort to resist; but so far we have managed reasonably well 
in the SW standards. I still feel optimistic.

It is important however to emphasise that ontology sharing is likely 
to be pretty simple stuff: its not the general AI problem. The 
optimism stems from the belief that even quite 'simple' stuff (from 
an AI perspective) is likely to have large utility in the actual 
world.

Pat

>
>Ugo


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes