Re: [ontolog-forum] Polysemy and Subjectivity in Ontolgies -- the HDBI E

["Gary H. Merrill" <gary.h.merrill@xxxxxxxxxxxxxx>]

Let me expand just a bit on part of Doug Foxvog's comment concerning applications of Cyc.

>> You mentioned Cyc and its microtheories for interpreting each meaning
>> of a word.  It doesn't seem to have worked, in that Cyc hasn't
>> gotten out of the public weal,
>> with contract work only, and only for research not for
>> practical applications.

> Cycorp has been much more focused on building the ontologies than on
> NLP.  It has been used for practical applications for over a decade,
> for example by a pharmaceutical company.

In fact, In the late 90s and into the early 2000s Cyc was used by two major pharmaceutical companies and a large health insurance company.  Although I used to manage the contract with Cycorp at GlaxoWellcome (and later GSK), I'm unsure of what the legal issues may be in disclosing details of some of this, but I feel able at this point to at least sketch the work that was done at GW and GSK.  I in fact never could get the other large pharma company to disclose to me the area or manner in which they were applying Cyc, but my impression was that at least for a while they were devoting serious effort to this over a period of years.

At GW, starting in appoximately 1996/1997 Cyc was used to implement what was known as the "GlaxoWellcome Thesaurus Manager", a web-delivered application for managing and (to some degree) integrating several large commercial thesauri in the medical realm, providing both enhanced search and thesaurus editing and integration capabilities on the basis of Cyc microtheories and the Cyc representation of concepts and their relations to terms.  I was quite heavily involved in this effort at virtually every level (from server configure and maintenance to debugging Cyc at the sublips level and consulting with the Information Management people who were the primary users and drivers of the application).  After I left GSK in 2001 (for a year at Novartis), responsibility for that was transitioned to someone else, and ultimately (just a few years ago) the Thesaurus Manager was decommissioned.  The major issue there was the lack of a clear vision (at GSK) for Thesaurus Manager use and the inability of the organization to support and fruitfully use such a tool in the changing corporate environment.  There was never a problem with Cyc or its applicability in this case.

In 1999/2000  I worked with Diraj Pathak (then in GW Bioinformatics) to create a (at that time) sophisticated system for use by bioinformatics in protein research.  This involved creating a microtheory specifically for that purpose (axiomatizing at least small portions of molecular biology and genetics), "importing" SwissProt into that microtheory, and providing enhanced search and knowledge discovery capabilities using Cyc's query and reasoning capabilities.  This was delivered as a fully functioning prototype in 1999, but it never progressed beyond that stage because of what is best referred to as "instability" in the bioinformatics department at that point in time.  Implementation of this system exposed a problem in using Cyc in such contexts for real-time knowledge exploration since its powerful reasoning capabilities tended to run into performance problems.  Had I continued with the project, however, I would have taken a slightly different approach and off-loaded some of the capabilities to a traditional RDB, fundamentally using Cyc to generate the database that would be capable of responding more quickly to interactive user needs.

I have a vague recollection that there were some minor and very high level descriptions of the Thesaurus Manager that were published or presented at conferences, and you can see it mentioned in some papers by folks at the University of Sheffield.  The work on the bioinformatics application was never published.

Based on my experience, I believe that the microtheory approach is a fundamentally sound one (in effect, it is a model and environment that captures a modal logic or counterpart theory representation in a computational context), and it provides great flexibility in representing knowledge and in inferencing.  I have had no experience with Cyc since 2000, but assume there have been some improvements of which I am unaware.  And it is exemplary in maintaining such distinctions as word and object, intension and extension, metalanguage and object language.

---

Gary H. Merrill
Ontolytics, LLC

+1 919.271.7259

_________________________________________________________________
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    (01)