Tim Berners-Lee is the originator of the World Wide Web and was listed byTime magazine as one of the 100 greatest minds of this century. His inspiring work on the Web set the stage for a world of changes in the way people do business, entertain themselves, exchange ideas, and socialize. In this podcast, Berners-Lee talks about his early history with the Web, opportunities and challenges at present, emerging technologies, and his current project: the semantic Web.
Podcast: Tim Berners-Lee
2006-08-22 Internet 5 Comments
I thought Al Gore was the inventer of the internet? Funny how wrong one can be inanit?
the semantic web won’t work. here’s why:
* people mean different things when they use, possibly the same, words – there is no way of agreeing on this. the use of “schemas” or ontologies are mere attempts at agreeing on what you mean by terms, but the problem is that even the definitions of these ontologies is subject to the same open problem.
* to make the above scheme successful you have to “restrict” your problem space, or the representation space. this is what is done with common computer languages .. everyone agrees what ”
” means in HTML or what “break” means in C – well enough agree to a sufficient degree so as to make the excerise useful. this is not what the web is about. the web is successful precisely becuase we gae up on a schema-enforced information-directory approach and let people uncategorised content. enforcing any useful amount of restriction would defeat the web – you will end up with a library database – not the ideals of the internet.
* you can see the depth of the problem when you see the effort put in my search engines.
* crawling robots or agents which play with some natural language processing does not count as the semantic web!
the semanic web is nothing but a buzzword – and it has traction because at one point hardly any research grants were issued unless the proposals mentioned the semantic web! (that honour used to belong to “distributed”!)
* the problem is essentially the same as that being chipped away at by the AI people – and that, we know, is vastly more complex than technology promoters can appreciate.
example: apple. do we mean apple as in Apple Computer or apple as in apple fruit? what if i change my web sie to use the more specific Pink Lady apples – how is an agent going to know what this means? if i publish a schema that says that a Pink Lady is a specialisation of the more general apple – how is that agent going to even parse my schema. and then how is it going to decode and attach *meaning* to my operators within my schema? and so it goes on … and all this so that the job of searching for the best prices for a weekly grocery shopping can be automated? how is it done today – by restricted and pre-agreed schemas and languages – far from the ideals of a semantic web as you can get.