Friday, January 20, 2006
[361] ß parallel discussion in “national debate” bead thread
This is part of a discussion that
will be moved to a Wiki page soon.
The Second School of Semantic Science, on
founding
[133] ß Judith Rosen’s recent
communication
Communication from the Protégé community
Paul,
let me say that your considerations are always thoughtful and
interesting to read (at least, those
parts my knowledge allows me to understand), even when teasing or constrasting with my own vision. Unfortunately, I
have no time to reply as extensively as I would like to, so I just make some
remark.
You are probably right about the terminology namespace problem, but then why not to rename also
"natural numbers", "complex numbers", and so on? Ambiguity exists... I think that with
little effort we can clarify the
context in which the terms we are using assume a particular meaning: indeed, we don't need namespaces to convey
information in natural language.
Also, please note that the term "complexity" in the computer
science community is just a shorthand for "computational complexity",
where the adjective "computational" is intended to clarify that the
domain of discourse are *computer* programs and not natural systems. Computer
Science (more precisely, Information Theory) tells us when and how computers
could replace -and perhaps do better than- humans, and is not concerned in any
way with figuring out what to do, with or without computers, when this is not
the case.
Instead, as far as I understand, the "second school" vision
is much about how computers can help humans doing better than computers, or
doing what computers alone cannot do: that seems to be an ambitious discipline,
maybe entangled with many other disciplines, and certainly *using* computer
capabilities -and thus, Information Theory- as a tool, is that right?
But then, since it is your discipline that benefits from the tools offered
by CS and IT, and not vice versa, why the language used by the former should be
changed to accommodate the one used by the latter?
Best regards,
Andrea
I see that the feedback I was looking for is in your note, and I thank you.
We might ask others to make comments about your note. Your response is, may be, classical and a poll of computer scientists would have 70 - 80 % responding precisely as you did (i conjecture).
You seem to say that computer science has a privileged position in the use of terms like "complexity", "semantic", "intelligence", "understands" and related terms. You seem to be saying that if natural scientists feel uncomfortable with the meaning given to these terms by AI, for example, then the natural scientists should develop their own language and not object to what they feel is the hijacking of specific important language.
The problem is, that computer science is sold (marketed) aggressively to the larger world society, and to the science community using language that in the wider context leads to misrepresentation and to "false claims". These false claims are not always even understood as misleading by the PhD in computer science who feels that the simplification of the phrase "computational complexity" to "complexity" is entirely appropriate in the context of discussions about modeling biological expression (such as gene or cell expression).
The problem becomes more sever when the work of forums like this one seem interrupted by the objection of the National Scientist, who is trying to suggest, or even conjecture, that what is being sold cannot be delivered.
These scientists point to principled arguments like those made by Robert Rosen, or Sir Roger Penrose. They also point to the many specific failures in the history of expert systems, AI and now the SW. They need to be able to see how the rewards of SW technology can be obtained, and for THEM this means that they cannot be required to throw away THEIR scientific training.
The language used in the SW (First School of Semantic Science) is sometimes offensive to a Natural Scientist, because the natural scientist sees intelligence in a far different light. Natural intelligence is not fully understood by science, and so the presumption that computer science figured it out correctly long ago is not a pleasing one.
Language is then the key bridge to a collaboration that brings value to the computer science from the natural science. If computer science has a let them eat cake attitude, then we all is lost - including eventually the computer science community.
As you may know, I have proposed that 25% (.4 B per year) of all academic funding current going to computer science from the US federal government (1.5 B per year) be re-programmed to support a new discipline called "knowledge science".
http://www.bcngroup.org/area2/KSF/nationalProject.htm
This proposal forces the issue to become political.
In the current SW and AI discussions the "false claims" issue is not vetted, and the notion of viewpoint is more diminished within the over all community. The computer science dominates as if by an Emperor (and this is why Penrose titled his 1989 book, "The Emperor’s new mind".
The second school suggests that the language errors are errors made primarily by the computer science community, and so this is where the corrections to those errors should occur. But in the case that natural science is in error, we wish also to correct these errors.
I am suggesting that everyone benefits if the OWL retrieval of information encoded into a OWL information base be referred to as "retrieval" rather than "inference". The reason is that the retrieval term does not over sell what the computer science is doing, and thus does not lead to misunderstanding.