W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2010

Re: Property Path Profiles

From: Matt Perry <matthew.perry@oracle.com>
Date: Tue, 12 Jan 2010 09:23:15 -0500
Message-ID: <4B4C85D3.2030307@oracle.com>
To: Andy Seaborne <andy.seaborne@talis.com>
CC: W3C SPARQL Working Group <public-rdf-dawg@w3.org>
Hi Andy,

Your observation is correct that we can transform this complex path into 
3 simpler paths. We are trying to identify those paths that can be 
translated to a single CONNECT BY.

-Matt

Andy Seaborne wrote:
> I don't understand the restriction to one unbounded path component.
>
> Looking at:
>
> { :v1 :myProp1*/:myProp2+/:myProp3? :v2 }
>
> It is the same as:
>
> { :v1 :myProp1* ?x1 .
>   ?x2 :myProp2+ ?x3 .
>   ?x3 :myProp3? :v2 }
>
> Matt - What is the key difference between the supported and 
> unsupported examples?
>
>     Andy
>
>
> On 11/01/2010 7:12 PM, Matt Perry wrote:
>> Hi Ivan,
>>
>> Basically, combinations of complex paths are not supported.
>>
>> For example, { :v1 :myProp1* :v2 }, { :v1 :myProp2+ :v2 }, and { :v1
>> :myProp3? :v2 } are all supported, but { :v1
>> :myProp1*/:myProp2+/:myProp3? :v2 } is not supported.
>>
>> Hope this helps,
>> -Matt
>>
>> Ivan Herman wrote:
>>> Matt,
>>>
>>> to make things more understandable (for me:-) can you summarize what 
>>> are
>>> the features in the current property path document[1] that are _not_
>>> covered? Ie, what do we give up if we use such profile(s)?
>>>
>>> Thanks for your help
>>>
>>> Ivan
>>>
>>> [1]http://www.w3.org/2009/sparql/docs/property-paths/Overview.xml
>>>
>>> On 2010-1-11 19:25 , Matt Perry wrote:
>>>
>>>> Hi,
>>>>
>>>> During the last TC, I mentioned the possibility of a Property Paths
>>>> profile that identifies a subset of property path queries that can be
>>>> expressed with SQL. Such a profile would make it easy for triple 
>>>> stores
>>>> implemented on top of relational databases to identify the set of
>>>> property path queries that they "natively" support.  The purpose of 
>>>> this
>>>> email is to start a discussion about the possibility of property path
>>>> profiles.
>>>>
>>>> The grammars below show two possible fragments that we have 
>>>> identified.
>>>> The first grammar is for SQL + CONNECT BY (Oracle) and the second 
>>>> is for
>>>> PLAIN SQL.
>>>>
>>>> CONNECT BY:
>>>>
>>>> ALT  ->  URI | URI|ALT
>>>> SEQ  ->  URI | URI/SEQ
>>>> Elem ->  URI | SEQ   | ALT   | ^URI
>>>> COMP ->  URI | Elem* | Elem+ | Elem{n,m} | Elem?
>>>> TOP  ->  URI | COMP  | ALT   | SEQ       | ^URI
>>>>
>>>> PLAIN SQL:
>>>>
>>>> ALT  ->  URI | URI|ALT
>>>> SEQ  ->  URI | URI/SEQ
>>>> Elem ->  URI | SEQ       | ALT   | ^URI
>>>> COMP ->  URI | Elem{n,m} | Elem?
>>>> TOP  ->  URI | COMP      | ALT   | SEQ   | ^URI
>>>>
>>>> Thanks,
>>>> Matt
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>> ______________________________________________________________________
>> This email has been scanned by the MessageLabs Email Security System.
>> For more information please visit http://www.messagelabs.com/email
>> ______________________________________________________________________
>
Received on Tuesday, 12 January 2010 14:24:13 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:41 GMT