[Top] [All Lists]

Re: [ontolog-forum] Fwd: Re: Using controlled natural languages for onto

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Sat, 12 Mar 2011 22:56:48 -0500
Message-id: <4D7C4080.80700@xxxxxxxxxxx>
Ferenc and Rich,    (01)

> ... the only marketable product is the upper ontology for which you should
> be making the case, because once that is set, the rest will be done by the
> domain engineers.    (02)

The point I have been repeating in different ways is that there is
no such thing as an ideal upper level.  Many people have reached
that conclusion after working very hard to find one.    (03)

Furthermore, even if somebody found some stone tablets with the perfect
upper level carved on them, nobody would buy it because they don't
have any tools that would use it.    (04)

In the 1990s, Doug Lenat said that the upper level wasn't as important
as the middle and lower levels.  He discovered that point the hard way,
namely by spending 10 years and quite a few million dollars.    (05)

I was also trying to develop an upper level around that time, but I
became convinced that Doug was at least partially right.  In my 2000
book, I presented an upper level -- but I clearly stated that it was
an example of "an" upper level, not "the" upper level.    (06)

For work on NLP, we have found that lexical resources with very few
axioms are the most important things you need.  The NLP gang from PARC
who went to PowerSet, which was bought out by Microsoft, discovered the
same point.  They had access to the full Cyc system.  But they didn't
find any use for the Cyc axioms.  They just used the bare hierarchy.    (07)

The BFO upper level is another example.  Just look at it -- no axioms
or detailed definitions.  Just a type hierarchy with some English
comments that look like definitions, but they're useless for any kind
of reasoning.  The only axioms are statements that say one undefined
category is disjoint from another undefined category.  Big deal!    (08)

> If you shrink-wrapped [the VivoMind software], who would use it,
> for what purpose, and in what environment?  Who would buy it,
> and how would they keep cash flowing in its usually convoluted
> eddies of markets and pain centers?    (09)

Probably nobody.  That's why we're not selling it.    (010)

> What will it cost to do that ASMOP, schedule to complete,
> earned value, etc.    (011)

Probably quite a few millions.  But it's not clear that it would
earn much money even if it were available right now.  Notice that
IBM designed Eclipse as a tool to compete with Microsoft Studio.
They couldn't sell it, so they gave it away.  IBM also developed
UIMA, but they gave it away to the Apache Foundation.  Tools are
very hard to sell.  People want applications.    (012)

What we are doing is working on contracts that pay the rent
and enable us to continue developing the tools and techniques.
At the moment, we're improving the tools so that we can
develop better applications more quickly.  That is
a business model for a small research company.    (013)

For the longer term, we are always looking at what kinds of
applications might be promising.  I haven't noticed that Bing
is using PowerSet technology yet.  And people aren't likely
to buy supercomputers that play Jeopardy.    (014)

John    (015)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (016)

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