- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Fri, 10 Sep 2004 11:33:58 -0700
- To: Sandro Hawke <sandro@w3.org>
- Cc: W3C TAG <www-tag@w3.org>
> Also, isn't the fatal flaw to "do nothing" the fact that you can't > possible tell whether > > <http://www.amazon.com/exec/obidos/tg/detail/-/006251587X> > dc:author "Tim Berners-Lee" > > is true or not? It's true if you take that to be the URI of a book, > and false if you take it to be the URI of a web page. No, unless you find somebody has died due to an ambiguous assertion. The subject is ambiguous because you seem to think of it as a web page (the collection of resource representations that result from an initial GET on the identified resource plus its subsidiary resources that are also retrieved, depending on the user agent configuration, until an application steady-state is achieved and the "user" can view the resulting page), whereas HTTP considers it to be just an identifier for the purpose of one request. The predicate is ambiguous because dc:creator is a catch-all metadata item that is not targeted to any singular aspect of creation nor limited to any specific type of object. The object is ambiguous because everyone knows that string literals don't author, so we just assume it is a name and that the name is somehow unique to the individual who is supposedly a creator of something vaguely identified as the subject. There is no way to make the above unambiguous by making false assumptions about the nature of the "http" scheme. If you want unambiguous assertions, you need to provide a great deal of elaboration and specificity to every term, just like any other KR system (and human language as well). Introducing a new syntax cannot solve fundamental AI problems. ....Roy
Received on Friday, 10 September 2004 18:34:31 UTC