Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

SUO: Re: Industry takeover




Adam,

I agree that we are at an impasse:

 > I'm afraid I disagree and I feel like
 > we're at an impasse again.  I keep asking for
 > concrete examples, proofs in logic, or some
 > evidence and you respond with English discussion.

What you are asking for comes at the end of
a project, not at the beginning.  You are asking
us to accept your faith in SUMO, which is supported
only by some vague notion (expressed in English)
that it might be useful for some commercial
applications.  We have no proof of that claim.

 > What are the claims you feel that have been made
 > about SUMO that you feel have not been justified?
 > I'd be happy to respond and maybe the Cycorp folks
 > could respond to any questions you have about Cyc.

Following are some questions that I would like to
have answered about any proposed upper-level ontology:

 1. Has it been implemented and used successfully in
    applications that are deployed and in continuous
    use (i.e., not merely research projects)?  For how
    long?  Has it made (or saved) money for the companies
    or agencies that use it?  How much?
   
 2. How has it been used in conjunction with other
    ontologies (including terminological ontologies
    such as WordNet, Mesh, UMLS, etc.)?  In research
    projects or in deployed applications in continuous
    use?  How have the categories of other ontologies
    been mapped to it?  Was the mapping proved consistent?

 3. Is there a proof of the consistency of the ontology
    as a whole (i.e. by a theorem prover that would
    have been able to detect an inconsistency, but
    terminated in a finite time without detecting any)?

These are questions I would like to see answered for
both OpenCyc and SUMO.  

There are other claims that I also find incredulous,
such as the notion that a monolithic ontology is
desirable.  Cyc gave up that idea a dozen years ago,
but it has still been claimed as a desirable goal for SUMO.

And finally, I see no justification for not collaborating
with OpenCyc and IFF.  If you want to have total control
over SUMO as an independent project, then do it yourself
at Teknowledge.  If you don't want to collaborate, then
don't ask a committee to rubber-stamp your project.

John