Re: typical distribution of arities?

At 10:52 AM -0500 7/19/07, Pat Hayes wrote:
> There are two rather different questions here.  One is statistical.

That is the thrust of my question.


> The other is about possibility, about what CAN be expressed. There is
> no need to do a survey for the second: the answer is that all one
> needs are unary and binary relations. In fact, all one really needs
> are binary relations. This result has been known in its essentials
> for about a century now.

Understood, but the results may not be convenient for either humans or
programs to deal with.  So, most KR systems support higher arities.
That being the case, I'd like to know how many of the higher arities
actually get used in practice.  My intuition is that the usage falls
off pretty quickly as N rises, but I'd like to see statistics.

-r

P.S.  FWIW, RDF looks to me like a two-address assembly language
      for knowledge representation.
-- 
http://www.cfcl.com/rdm            Rich Morin
http://www.cfcl.com/rdm/resume     [email protected]
http://www.cfcl.com/rdm/weblog     +1 650-873-7841

Technical editing and writing, programming, and web development