# Re: Propositions

sowa <sowa@turing.pacss.binghamton.edu>
Date: Wed, 16 Feb 94 06:17:12 EST
From: sowa <sowa@turing.pacss.binghamton.edu>
Message-id: <9402161117.AA15498@turing.pacss.binghamton.edu>
To: cg@cs.umn.edu, interlingua@ISI.EDU, schubert@cs.rochester.edu,
sowa@turing.pacss.binghamton.edu
Subject: Re: Propositions

Len,
Yes, the problem with variable names is indeed vexing. In fact,
that is one of the main reasons why I prefer to dispense with
variable names completely. The issue simply does not arise with
existential graphs or conceptual graphs. But then you have other
issues, such as defining a lexicographic sorting of subgraphs whose
nodes are unnamed.
That is where the Levinson-Ellis algorithms for maintaining
generalization hierarchies may be important -- they sort the
graphs in a hierarchy that is determined by structure alone,
without considering the names of the variables that link them.
For the kinds of structures that arise in a typed logic, the
type names are very important for reducing the amount of testing
in comparing graphs. But types are a semantically significant
category, unlike names, which are purely connective.
John