Re: Defining property paths

On 12/03/12 21:09, Birte Glimm wrote:
>> The nodes recorded are just those the subpath ends on.  So if the subpath is
>> :p/:q then the overall path can loop back through the gap between a :p and
>> :q.  This makes sense to me - its working over a (non-RDF) graph induced by
>> only considering :p/:q as links.
>
> I am not quite sure I get this.

Example:

:x1 :a :x2 .
:x2 :b :x3 .
:x3 :c :x4 .

:x4 :a :x3 .
:x3 :b :x5 .
:x5 :c :x6 .


Path
:x1 (:a/:b/:c)+ ?Z

which goes :x1 -> :x4 -> :x6
i.e. goes through :x2 twice (in some sense) for different reasons.

If you link about it in the graph derived by taking :a/:b/:c links:

:x1 ":a/:b/:c" :x4 .
:x4 ":a/:b/:c" :x6 .

it makes sense and has no repeated node.

 Andy

Received on Tuesday, 13 March 2012 13:18:35 UTC