On Feb 18, 2009, at 5:53 PM, Patrick Cassidy wrote:
Can you point me to an application of OpenCyc (one that I can view and
test with a browser or test after downloading and installing for free) that
gives a good example of interoperability among programs created by two or
more separate development groups? Where can I find the translator that
takes a KB developed using SUMO and translates it into OpenCyc assertions?
Is there a web site where we can test out the Cyc language understanding
function? Can we modify modules of that program to see if a different
approach will result in superior performance? If a commercial company wants
to use the Cyc reasoning engine can they do it without paying royalties? Is
there a set of separately developed databases integrated via OpenCyc that
can be globally queried by either a web browser, or by downloading and
installing? Are there a collaborative open-source projects to create
applications using OpenCyc, open to contributions from any interested party,
with the results freely usable?
No, because Cyc is a commercial project. And it is that because that was the only way that Doug could finance it over the long (20-year and counting) timespan he knew would be required to create it. Unlike you, he did not spend his time emailing to open forums about the need to find millions of dollars from somewhere: he set out, in an enterpreneurial spirit, to actually get the funding, and he got it, and did the work. Now the work is done, it would be cheaper to buy the product from his company than to re-do hundreds of man-years effort with public money.
Under the circumstances, to
suggest spending millions of public dollars to re-create something
functionally indistinguishable from it seems quixotic at best, and
irresponsible at worst.
If the existing OpenCyc cannot do all of the above (and it can't) then it is
**not** "functionally indistinguishable" from the FO and associated programs
that would result from this project.
You have not demonstrated, or even AFAIK convincingly argued, that the FO will be able do any of these things, either. I fail to see how an ontology of any kind
can achieve interoperability or translation all by itself. The FO as you have described it will be an ontology, not a universal semantic Swiss Army Knife.
I know that the Cyc is technically
*capable* of doing those things, if some open community of users were to
adopt it for that purpose and put effort into developing it in that
direction, but no such community now exists. I too have a copy of the full
Cyc, but I have absolutely no incentive to try to develop any applications
for it because I know that any interesting application would have to be
developed by a rather large community of developers working together, and no
such community now exists for Cyc as far as I can tell. Developing a modest
application on my own would not in any way advance the goal of semantic
If the application demonstrated such interoperability, of course it would. Why not just set out to do that? The Army will probably support you for a couple of years, which should be time enough for a convincing demo.
That is one of the lessons we should have learned from
the experience of the past seven years, since the first OpenCyc became
available. In the meantime, 700 billion dollars and counting of economic
inefficiency are down the drain.
If there is an open-source open-participation project using OpenCyc that is
aimed at developing a language-understanding program, please point me to it.
If not, please recognize and admit that there are things that an FO
consortium could do that are not now being done, and are not likely to be
done, by a group of Cyc users in the foreseeable future without funding of
some project to that purpose.
There are of course things not being done that could be done. I am not convinced, not will I admit, that your proposed FO consortium would be able to do anything useful at all. It would be too large, too loosely organized, and would not have a clear aim or purpose. It would be a committee designing an abstract mammal, not even knowing whether it was trying to be a horse or an elephant. In practice, the key decisions will be made by small inner active subgroups, and those outside the active subgroup will not be happy with all aspects of the resulting design. Eventually, the group will either adopt one alternative by fiat or voting, or else it will fracture or become unworkable. The first option will produce a second Cyc; the second option will result simply in paralysis.
I know you don't agree, but that is my own prediction of what is the likeliest outcome, which is why I don't recognize that there are things the FO could do that Cyc can't.
[PC] , and does not now advertise an ability to translate among
[PH] Fair enough. BUt if this is supposed to the raison d'etre for the FO,
then demonstration of how to achieve such translations and
interoperabilty in a few nontrivial cases would be a good place to
begin. We managed to do this in the IKRIS project, for example, to
limited but measurable extent. Cyc was one of the targets, in fact. I
bet that a well-designed and convincing proposal to achieve multi-path
interoperability between a large number (say, 20 or so ) existing
formalisms and ontologies and notations could in fact attract
substantial funding right now, as many government agencies have
identified this as a major problem area.
If there is such a proposal in the works with the goal of creating
open-source Cyc-powered programs open to participation from any source, it
might be a good substitute for the FO project if it has a natural-language
understanding component, and I would be happy to participate, but I am
unaware of it. I have read at least some of the IKRIS reports, and know
that you also have an interest in (and optimism about) enabling translation
among alternative representations. So then why don't you organize such a
project if it is likely to attract funding?
Because I believe it is way too early to be trying to normalize such interoperability, and that the best, perhaps the only, way to achieve it is via some version of the semantic web. Maybe not using current SWeb technology, but based on the overall concept of distributed syndication and open publication of ontologies. No centrally administered project, no matter how large, can hope to achieve the potential scope of a Web-based effort. So I am bending my small efforts largely in the direction of trying to make the SWeb more usable and more widely used.
I will be happy to help in any
way I can.
Given the large economic costs of the delay involved in taking a very slow
incremental approach with many isolated projects, I think we should have
some sense of urgency in organizing such an open collaborative project,
whatever you call it. But the resulting products do need to be fully open
source and easy to test by casual inspection, in order to encourage new
users. Can you provide us with your counter-proposal?
I am not the one making proposals here, you might have noticed. I don't think any such grand project has a snowball's chance in hell of succeeding. I don't even think that success is well-defined for a project like this. So no, I am not making any proposals along these lines.
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
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
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
IHMC (850)434 8903 or (650)494 3973
40 South Alcaniz St. (850)202 4416 office
Pensacola (850)202 4440 fax
FL 32502 (850)291 0667 mobile