Re: comment on the fragment document: (inverse) functional and DL-Lite

On Thu, 6 Mar 2008, Ian Horrocks wrote:
>
> Actually, the primary idea of DL-Lite is that it has logspace data 
> complexity, which means that query answering can be implemented using 
> standard (relational) DB technology. In particular, a conjunctive query 
> against a DL-Lite ontology can be re-written as an SQL query against a 
> relational DB containing instance data.

I agree, the goal of DL-Lite is: being able to answer conjunctive
queries under a OWL-DL style open world semantics by storing the ABox
in a relational DB, rewriting the query, and then using standard RDB
query answering. This can be viewed as backward chaining and has the
advantage that off-the-shelf (completely unmodified) DB technology
can be used to answer conjunctive queries under an OWL-DL semantics.

A slight correction though: logspace data complexity does not
necessarily mean that this can be done (at least such a general result
is unknown). However, the following is true: non-logspace data
complexity means that this can *not* be done.

greetings,
 		Carsten

> Ian
>
>
> On 6 Mar 2008, at 19:23, Jim Hendler wrote:
>
>> 
>> I thought the primary idea of DL-Lite was that it would provide primary 
>> database functionality -- how can it do that without keys?
>> 
>> 
>> On Mar 6, 2008, at 6:18 AM, Bijan Parsia wrote:
>> 
>>> 
>>> On 6 Mar 2008, at 11:04, Ivan Herman wrote:
>>> 
>>>> Boris, Bernardo,
>>>> 
>>>> I went through
>>>> 
>>>> http://www.w3.org/2007/OWL/wiki/Fragments_Proposal
>>>> 
>>>> again today. One thing that I may have missed: I tried to see if I can 
>>>> use (inverse)functional properties for DL-Lite or not. I did not find any 
>>>> reference to those neither in 3.1 nor in 3.2. Again, I may have missed 
>>>> something...
>>> 
>>> Let's see if I can discern from the text the situation. (As a test of the 
>>> spec.)
>>> 
>>> In section 3:
>>> 	http://www.w3.org/2007/OWL/wiki/Fragments_Proposal#DL-Lite
>>>
>>> 	"""Several variants of DL-Lite have been described in the literature. 
>>> The variant presented here is called DL-LiteR since it allows for property 
>>> inclusion axioms; it therefore contains the intersection between RDFS and 
>>> OWL 1.1 DL. Other variants trade property inclusion axioms for 
>>> functionality and inverse-functionality of object properties."""
>>> 
>>> I think this is clear that functionality and inverse functionality of 
>>> *object* properties are forbidden.
>>> 
>>> Actually ,the rest of the sections are quiet about data properties 
>>> altogether. Which would mean that data properties are forbidden in this 
>>> variant. Which means that it's not really the intersection of RDFS and OWL 
>>> 1.1 DL?
>>> 
>>> I do think that if we make this DL Lite not have data properties, the text 
>>> should call that out (e.g., in the list of missing features). OTOH, I 
>>> think we should allow data properties ;) I would think it would be ok to 
>>> trade datasubproperties for keys (from a user pov)...I don't know if that 
>>> would be ok from the logic/impelmentation pov off the top of my had (while 
>>> retaining object subproperties).
>>> 
>>> Cheers,
>>> Bijan.
>>> 
>>> 
>>> 
>> 
>> "If we knew what we were doing, it wouldn't be called research, would it?." 
>> - Albert Einstein
>> 
>> Prof James Hendler 
>> http://www.cs.rpi.edu/~hendler
>> Tetherless World Constellation Chair
>> Computer Science Dept
>> Rensselaer Polytechnic Institute, Troy NY 12180
>> 
>> 
>> 
>> 
>> 
>
>

--
*      Carsten Lutz, Institut f"ur Theoretische Informatik, TU Dresden       *
*     Office phone:++49 351 46339171   mailto:lutz@tcs.inf.tu-dresden.de     *

Received on Friday, 7 March 2008 08:45:57 UTC