- From: Tim Berners-Lee <timbl@w3.org>
- Date: Tue, 7 Jan 2003 10:38:19 -0500
- To: www-style@w3.org
- Cc: Shelby Moore <shelby@coolpage.com>
(First of all let me start with a warning that the word "semantics" may be used and often has been used in a very different way in markup language field to the field of knowledge representation, and that the Semantic Web uses it in a specific way, so beware of misunderstandings there.) Shelby wrote, in <http://lists.w3.org/Archives/Public/www-style/2003Jan/0059.html>, > Message-Id: <4.1.20030103143924.018f2700(null)> > Date: Fri, 03 Jan 2003 14:47:24 -0600 > To: Ian Hickson <ian@hixie.ch> > From: Shelby Moore <shelby@coolpage.com> > Cc: www-style@w3.org > Subject: Re: XBL is (mostly) W3C redundant, and CSS is wrong W3C layer > for semantic behavior *markup > > > I am searching the w3c.org to see if I can find a definition of > "semantics" > that relates to specifications. > > So far no concise definition found. > > I did find something related to semantics from Tim Berners-Lee on > goals of > Semantic Web: > > http://www.w3.org/DesignIssues/RDFnot.html#Knowledge > > "Each field had made certain ___CENTRALIST__ assumptions -- if not in > the > philosophy, then in the implementations, which __PREVENTED__ them from > spreading globally...We __REMOVE__ the __CENTRALIZED__ concepts of > absolute > truth, total knowledge, and total provability..." > > http://www.w3.org/DesignIssues/RDFnot.html#Surely > > "The __PROBLEM___ with all such systems was that they were > conceptually or > physically __CENTRALIZED__. They required link __GLOBAL__ > __CONSISTENCY___." > > > So what Tim Berners-Lee is saying is same as what I am saying, which is > that you can not centralize semantics. > > Ian, there are the references you requested to refute your concept of > semantics. > Shelby, I think you are quoting me out of context here, and possibly inappropriately. My text above argues about a need for a decentralized web of concepts for representing global knowledge. However, while one cannot insist on a single tree of concepts and global standardization for everything, such standardization is in most cases a good thing where it can happen. In particular, the common infrastructure relies on common concepts, as defined by specifications. This is really important. You use my words about the impossibility of defining global knowledge to argue against the authority of the specification, and the authority of the specification is very important. I have my keynote at WWW in 2002 almost entirely on that point. I have written about it in the draft <http://www.w3.org/DesignIssues/Stack.html> To summarize my view, yes - for natural language usage defines meaning, but for technical specifications we have a a social convention that terms are used in accordance with meanings defied in specifications. Other usage is wrong. It happens, but is wrong and should not be condoned. There are corner cases where the specification system doesn't work but in general it is a very important system and should not be denigrated. Interoperability relies on usage according to the specification. Specifically now that commerce occurs on the net and and its associated legal wrangling occurs about the meaning of what has been said, we mustn't allow people to try to philosophize their way out of social responsibilities. For example, spammers cannot claim (as one said to me on the phone) "Oh, the RFCS aren't law, so I can put anything in the From: field I like". Tim BL > I will look for more... > > -Shelby Moore
Received on Tuesday, 7 January 2003 10:38:41 UTC