Re: wffs vs graphs

Don Dwiggins <dwig@markv.com>

Mail folder: Interlingua Mail
Next message: sowa: "Back to propositions"
Previous message: phayes@cs.uiuc.edu: "Re: wffs vs graphs"
In-reply-to: schubert@cs.rochester.edu: "wffs vs graphs"

Subject: Re: wffs vs graphs
To: cg@cs.umn.edu
Date: Fri, 25 Feb 1994 15:21:16 -0800 (PST)
From: Don Dwiggins <dwig@markv.com>
Cc: boley@dfki.uni-kl.de, fritz@rodin.wustl.edu, interlingua@ISI.EDU,
        phayes@cs.uiuc.edu, schubert@cs.rochester.edu
In-reply-to: <199402251846.NAA15811@cherry.cs.rochester.edu> from "schubert@cs.rochester.edu" at Feb 25, 94 01:46:14 pm
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Content-Length: 1076      
Message-id: <9402251521.aa17505@hermix.markv.com>

schubert@cs.rochester.edu writes:
> 2. A graphical representation of a formula could be exponentially
>    more compact than a string representation, by avoiding subexpression
>    repetition. 
> 
> The latter is an interesting point. But as I noted in my article in
> John's book, one can augment standard logical syntax with subexpression
> names, getting something even closer to network syntax. With these
> names, the wff-syntax can completely avoid subexpression repetition,
> and also allows cycles (e.g., Liar sentences, illustrated in the paper).

This sounds like macros in programming languages; can the subexpression
names be given parameters?

Speaking of programming languages, the work on abstract syntax trees in that
field might provide some useful ideas for this discussion.  Programming
languages (at least the conventional ones) also have to worry about variable
names and their scopes and bindings, and some of the optimization algorithms
involve detecting common subexpressions.

Don Dwiggins				"Solvitur Ambulando"
Mark V Systems, Inc.
dwig@markv.com