Requiring nodes to be blank Re: An alternative RDF

On 2014-07 -10, at 14:17, Victor Porton <porton@narod.ru> wrote:

> 10.07.2014, 16:05, "Michael Brunnbauer" <brunni@netestate.de>:
>> Hello Victor,
>> 
>> when using RDF, you cannot decree that certain entities have to be described
>> using blank nodes. RDF users are free to choose between URIs or blank nodes and
>> your app should be able to cope with both.
> 
> My conclusion from this is that the RDF specification should be changed, so that an application would be able to demand that certain nodes are blank.
> 
> And despite of that it is not already in the RDF standard, it seems that most of implementation already provide support for this: Using an RDF library I can tell which nodes are blank and which are not.
> 
> Again: I demand to change the RDF specification. The rationale: Users should be able to construct an RDF file in such a way that loading more RDF files would not break its consistency. It is important.

Victor,

You are saying you want to force a structure like

       <#office>   :manager [ :name "Bob";    ssn 123456789;   :dob  "1966-01-01"  ] .

and not 

     <#office>   :manager  <#Bob>.
     <#Bob>  :name "Bob";    ssn 123456789;   :dob  "1966-01-01"  .

because you will be safe from encountering data from someone like

     <#Bob>  :dob  "1966-05-06"  .

which would lead you to conclude in total

     <#office>   :manager  <#Bob>.
     <#Bob>  :name "Bob";    :ssn 123456789;   :dob  "1966-01-01" ,  "1966-05-06"  .

which is obviously inconsistent to anyone who knows :dob is functional.

But if you have the blank node

       <#office>   :manager [ :name "Bob";    ssn 123456789;   :dob  "1966-01-01"  ] .

and someone also provides a graph

	[ :ssn 123456789 ]  :dob  "1966-01-01"  .

then your total data is anyway inconsistent to anyone who knows that :ssn is inverse functional.
Your problem is not the shape of the graph, it is you have to decide which to believe.
You have incorrect data.

Depending on your application, your choice will mess up statistics in a study in a very minor way, or someone will be denied a job, or a plane will crash.  RDF won't tell you what to do in general.


timbl