Re: SUO: RE: Topic Maps
Pat,
I completely agree with you:
>... Novakian Cmaps
>are widely used in classsroom applications and are apparently of
>considerable use in aiding inter-personal collaboration (other folk
>in this Institute have been developing Cmap interactive educational
>software for years, with Novak as an external consultant; you can
>download it free from http://cmap.coginst.uwf.edu/ ) but they are
>most definitely not a KR formalism. They have no clear semantics, or
>indeed even a clear logical syntax.
Unfortunately, I often run into people who can't tell the
difference, and they try to use CGs like concept maps and
vice-versa. I mentioned them in my CS book as a possibly
useful first-step (in the same sense that an outline may be
useful in writing a book or paper).
>I am currently working on a Darpa-funded effort to build software to
>allow 'subject-matter experts' (read: people who don't know logic) to
>input Cmaps which will get translated into a suitable KR formalism,
>and it is currently unclear whether this will turn out to be easier
>than translating natural language into KR or more difficult:
>certainly it is a research problem at about the same scale of
>difficulty.
I agree. The difference between CGs and Cmaps is very much
like the difference between unrestricted NL and a controlled
language, such as ACE.
Suggestion: Rather than translating already developed Cmaps
to logic, it might be better to design interactive tools that
help the person who is drawing the Cmaps to include the
necessary info that is needed to generate the output logic.
(But I admit that designing good tools to do that is very
much a research problem in human factors or HCI, as they say.)
John Sowa