Re: AW: [ontolog-forum] Current Semantic Web Layer Cake

On 6 Aug 2007, at 13:01, <tim.glover@bt.com>  wrote:

>
> Henry,
>
> Good answers, thank you for taking the time to answer me. You have  
> made
> many detailed points and I won't waste your time by going through them
> all.
>
> I will just repeat the thrust of my main point which is that data
> integration is difficult because people are inconsistent in the way  
> they
> use terms, and using URIs cannot prevent that, dereferencing
> notwithstanding. The difficulties arise when people are not aware that
> they are using the same term for different things.  But I'll concede
> that perhaps using URIs can help :)

good :-)


> Where the "rubber hits the road" is with inconsistency. If we believe
> that URI's will solve the problem of inconsistent use of terms (read,
> data integration problems), why all the stuff on the SW discussion  
> lists
> about dealing with inconsistency?

Ah. That is where you have to look at named graphs. People can be,  
and will be inconsistent.
People will lie and cheat.

see http://blogs.sun.com/bblfish/entry/beatnik_change_your_mind

The solution are named graphs. The client needs to keep track of  
where he got the information
from. And merge and unmerge depending on the trust he has in the  
sources of the information.

This is what we do in real life. It is complex, and that is why the  
first semweb applications have
to be simple. The tools to work with Named Graphs have only be  
appearing very recently.

I suggest that the first ones, like my prototype Beatnik Address  
Book, need to make it easy for users
to specify trust that they have in files, see the consequences of  
believing something, be able to
change trust values, etc... This will then help crate feedback to  
producers of defective rdf, and sidelines liars.

Henry


> Best wishes,
>
> Tim.

Received on Monday, 6 August 2007 11:22:10 UTC