[Date Prev] [Date Next] [Thread Prev] [Thread Next] Indexes: Main | Date | Thread | Author

[ba-ohs-talk] Fwd: CG: Re: Program Semantics


A particularly interesting discussion follows...    (01)

>From: "John F. Sowa" <sowa@bestweb.net>
>
>Bill,
>
>In the interest of contributing to the "pragmatic" side of the SUO
>effort, I'd like to mention some examples that illustrate my point
>that the same kinds of category mistakes that create ambiguities
>in natural language also create problems for reasoning systems.
>
>The first is taken from the book _Ahab's Trade:  The Saga of South
>Seas Whaling_ by Granville Allen Mawer, p. 177:
>
>    Medical arrangements were primitive.  British and French whalers
>    by law carried doctors, but Americans were made of sterner stuff.
>    One of the deductions from the seaman's lay was a contribution to
>    the medical chest, which was usually of good quality, but if he
>    was in need of something from it he was at the mercy of the captain's
>    skill as a pharmacist.  The ingredients in the chest were labeled
>    with numbers and there were instructions about what to combine with
>    which to treat that.  One is reluctant to dismiss as altogether
>    apocryphal the story of the captain who, having run out of number 9,
>    dispensed a mixture of 4 and 5.
>
>Category mistake:  Operations on the category of numbers when used
>for counting and measuring cannot be applied to any category for which
>the nubmers are used as identifiers.
>
>Anyone who has any experience in debugging computer programs has
>probably seen other category mistakes caused by applying operations
>appropriate to numbers to other kinds of entities. One of my favorites
>is the following response from a database question-answering system:
>
>   Q:  What is the largest state in the USA?
>
>   A:  Wyoming.
>
>To answer questions of the form "largest X", the program would sort
>the X's and return the last one.  But numbers are the only category
>of entities whose names are directly related to their sizes.
>
>Another example from _Ahab's Trade_, p. 192:
>
>     ... a chief named Vendovi [from Fiji], who had been responsible
>     for the killing and eating of a number of American seamen,
>     was surrendered, but again only after hostages had been seized.
>     He readily confessed and showed no contrition for what he clearly
>     regarded as normal behavior.  His followers were even aggrieved
>     that the seamen, especially a Negro, had tasted of tobacco.
>
>Imagine the dire consequences to any sailor who might have thought
>that the Fijians were looking for seamen with good taste.
>
>Examples like these can be multiplied endlessly.  They all show that
>the difficulties with natural language have nothing to do with any
>peculiarities of NL itself, and they have everything to do with
>the same kinds of ontology issues that create problems for computer
>databases and reasoning programs.
>
>Bottom line:  If you want to understand ontology, the best source
>of examples, problems, and solutions comes from the study of
>natural language semantics.
>
>John Sowa
>
>========================================================================
>To post a message, send mail to cg@cs.uah.edu.
>To unsubscribe, send mail to majordomo@cs.uah.edu with the command
>'unsubscribe cg' in the message body.
>See http://www.virtual-earth.de/CG/cg-list/ for the mailing list archive
>See http://www.cs.uah.edu/~delugach/CG for the Conceptual Graph Home Page
>For help or administrative assistance, mail to owner-cg@cs.uah.edu    (02)