Re: Semantic of the language

I'm looking at the SPARQL spec, so take for example ORDER BY, which cannot
be represented in datalog (right?).

It states that "The ORDER BY clause establishes the order of a solution
sequence." [1] So once you get a solution sequence, the order is established
in post-processing. Therefore there are some things that don't have to be
explicitly in FOL.

[1] http://www.w3.org/TR/rdf-sparql-query/#modOrderBy

Juan Sequeda
+1-575-SEQ-UEDA
www.juansequeda.com


On Tue, Mar 16, 2010 at 7:25 AM, Lee Feigenbaum <lee@thefigtrees.net> wrote:

> On 3/16/2010 7:40 AM, Marcelo Arenas wrote:
>
>> On Tue, Mar 16, 2010 at 2:03 AM, Lee Feigenbaum<lee@thefigtrees.net>
>>  wrote:
>>
>>> On 3/15/2010 11:55 AM, Marcelo Arenas wrote:
>>>
>>>  + the ability to project expressions from queries rather than variables
>>>>> only
>>>>>
>>>>
>>>> I don't understand this functionality. Maybe Lee could gives us an
>>>> example of such a query.
>>>>
>>>
>>> In SPARQL 1.0, the only thing that can be projected from a query are
>>> variables:
>>>
>>> SELECT ?var1 ?var2 ?var3
>>> WHERE {
>>>  ... ?var1 ... ?var2 ... ?var3 ...
>>> }
>>>
>>> In SPARQL 1.1, any expression can be projected into a result set:
>>>
>>> SELECT ?var1 (?var2 + 2 AS ?sum) (fn:myCrazyFunction(?var3) AS ?crazy)
>>> WHERE {
>>>  ... ?var1 ... ?var2 ... ?var3 ...
>>> }
>>>
>>> Lee
>>>
>>
>> Thanks for the example, now the functionality is clear. Whether this
>> is expressible in FOL depends on the functions that are used. For
>> example, ?var1 + ?var2 AS ?sum is not expressible in FOL if + is not
>> part of the language; if + is not included as a built-in predicate (or
>> function) in non-recursive Datalog with safe negation, then the above
>> functionality of SPARQL 1.1 will not be expressible in the mapping
>> language.
>>
>
> I still haven't read the papers that Juan cited, but I'm guessing then that
> a similar caveat was already true about the expressivity of SPARQL 1.0,
> given that FILTERs can make use of an extensible set of functions?
>
> I suppose just about any route we go for the mapping language will involve
> having to pick out a core set of functions/operators that can be used in
> transforming vales...
>
> Lee
>
>
>> All the best,
>>
>> Marcelo
>>
>>