[topicmapmail] [ RDF and TMs ] stupid questions

[knud.moeller]

M-L Chung writes: 

> Hello everyone: 
> 
> I know maybe I'm asking stupid questions, but I really need your help.

Not stupid at all! 


> The last question is:
> I know we can define semantic relationships between topics/resources with TMs(or RDF).
> But as a user, I cannot realize the significant differences between the websites
> have been built with TMs(or RDF)  and those have not. 
> I cannot perceive their "power" of  knowledge representation or knowledge organization
> because we can also represent semantic relationships without using TMs(or RDF).
> ( I did not mean to offend anyone.)
> Could anyone tell me what's the advantage for a website(or digital library, portal...)
> building with TMs(or RDF) ?  (especially the advantage which couldn't be accomplished
> without TMs [or RDF] )

OK, I'll try to sum up my impression of the whole thing. The web as we know 
it - and in fact, any kind of collection of resources like texts, pictures, 
etc. - are easy enough to understand for humans. You can read the texts, 
look at the pictures, follow the links, fill in the forms. You can make 
sense of it. However, a machine can't. Machines can display stuff and 
transfer it. 

So, one of the basic ideas of RDF (and higher-level models like OWL or 
RDF/S, and to some degree also Topic Maps) is to give a formal description 
of knowledge. Once formalized, a machine can make (more) sense of it, and 
help you to find things (even though Google and co. are already pretty 
clever at that), organize things, etc. At some point machines might even be 
able to come up with information that wasn't explicitly there in the first 
place (by using a logical framework like Description Logics, on which OWL is 
based). 

Another, more practical advantage of using models like TMs or RDF is that 
you can describe some domain in a general way, not directly tied to specific 
webpages, etc. Instead, you have your domain model and produce things like 
webpages dynamically. That's what TM or RDF driven webportals do. That way, 
your model of the domain is much easier to manage and maintain. 

So yes, while a user might not immediately see the difference between a hard 
coded webportal or one that is based on some semantic model, it does have 
other advantages. And the hope is also that technologies like TMs or RDF, 
OWL, etc. might help to produce better quality webportals easier. 

With respect to RDF there are other things (distributed knowledge, 
reasoning, partial knowledge, etc. all related to the idea of the Semantic 
Web), but as far as I understand, these are not so relevant for TMs. 

I guess there are many other opinions on the subject, but these were just 
some quick thoughts I came up with. 

Cheers,
Knud