Re: fact checking for semantic reasoners

Do you mean 'is Fred a bat?'
If we know some class of knowledge about Fred thenhe possibly could be a bat. 
Then another class of information knowledge would confirm or disconfirm this. 
About this second class of knowledge, facts about Fred, there is talk of provenance. I cannot see how this can be achieved apart from in a limited and restricted way. 
When I stick my head out of the window I report that it is raining while you, somewhere else in the room take my word. 
What is the process of taking my word?
It isn't because you know me. If my report were of a more complex nature it may be because you know I am human. 
Nevertheless I think this means that reports from machines are intrinsically unreliable precisely because the human is masked by a machine process in one way or another. That is by anything from complex procedure to complex algorithm. With all the possibility of bias, cheating and gaming in-between. Precisely when you can't see if I stuck my head out of the window or not. 
This leads me to conclude that this sort of endeavor inevitably creates the ground for dishonesty and corrodes trust between people and human entities. This is the very serious downside of creating mechanisms that generate large scale accurate data. Those mechanisms can, and that means they will, be manipulated. I remain to be convinced that this can be avoided to any significant degree or that it is not the case.
Whether this is qualitatively different to media manipulation as it stands in TV and news print is a subject that should be high on people's agenda. 

Adam

Sent from my iPhone

On 29 Aug 2011, at 14:13, Paola Di Maio <paola.dimaio@gmail.com> wrote:

> Enrico
> 
> thank you-
> 
> I guess we are thinking of slightly differently about this, but
> I agree with your suggestion that what I am aiming for is 'global consistency'.
> 
>> 
>> are the consistency checks in ontologies validated by supported evidence, or how?
>> 
>> I suppose I am thinking of  the situation where there reasoning spans many ontologies with  conflicting axioms
> 
> 
> you see, new 'evidence' comes up all the time. not sure if ontologies (as far as I know, as we know them today) are capable of being updated at the frequency rate that new knowledge is generated/published
> 
> I suspect, if new evidence cannot be taken into account, 
> then the reasoning can be obsolete one minute from the next
> (the argument overlaps the dynamic ontologies forum)
>  
> 
> You are now mixing pears and oranges,
> 
> oh no,  I am simply talking about the fruit, its another class level altogether :-)
> 
>  
> 
>> would be nice to have a service as such
> 
> we do.
> 
> a standanle web service that can 'fetch evidence for assertion x' on wiki/dbpedia is what I am thinking of
> 
> pointers to existing services welcome
> 
> 
> 
> thanks
> 
> P