- From: Drew McDermott <drew.mcdermott@yale.edu>
- Date: Tue, 30 Dec 2003 11:42:16 -0500 (EST)
- To: public-sws-ig@w3.org
> [me]
> The conclusion I would draw is that typed languages are a good thing,
> for a variety of reasons:
>
> (a) They make syntax checking possible. In KIF and Prolog, there is
> virtually nothing to check. (I'm not sure exactly what I'm checking
> when I check an RDF/OWL file, but besides verifying that I never said
> '&foo;' when any idiot can see I meant 'foo:', I suppose I'm verifying
> that a property is not used on an object outside its domain, which is
> at least close to type checking.)
>
Bijan Parsia has pointed out (personal communication) that one way to
think about OWL and other DL-ish systems is that they generalize type
checking in order to get the advantages (efficiency, decidability) of
type checking algorithms while doing a broader class of inferences.
So a vote for type checking is in a way a vote for DLs, which is a
vote I never thought I'd cast.
--
-- Drew McDermott
Yale Computer Science Department
Received on Tuesday, 30 December 2003 11:42:31 UTC