- From: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk>
- Date: Wed, 5 Aug 2009 15:46:23 +0100
- To: Pascal Hitzler <hitzler@aifb.uni-karlsruhe.de>
- Cc: Jim Hendler <hendler@cs.rpi.edu>, Jie Bao <baojie@cs.rpi.edu>, Uli Sattler <sattler@cs.man.ac.uk>, Sandro Hawke <sandro@w3.org>, "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, public-owl-wg@w3.org
Great -- thanks! I substituted this diff for the one previously given in the draft response [1]. Hopefully we are now *really* ready to go with this response. Ian [1] http://www.w3.org/2007/OWL/wiki/LC3_Responses/GB1 On 5 Aug 2009, at 14:35, Pascal Hitzler wrote: > okay, I changed it into the wording you suggest below. > > Diff (to our original version): http://www.w3.org/2007/OWL/wiki/ > index.php?title=Primer&diff=25100&oldid=25085 > > Pascal. > > Jim Hendler wrote: >> umm, okay, but I think you may have gone overboard (up to you) -- >> you changed >>> By and large, different profiles can be distinguished >>> syntactically with there being inclusion relations between >>> various profiles. For example, OWL 2 DL can be seen as a >>> syntactic fragment of OWL 2 Full and OWL 2 QL is a syntactic >>> fragment of OWL 2 DL (and thus of OWL 2 Full). Each of the >>> profiles presented below is a (strict) syntactic subset of OWL >>> DL, but none of these profiles is a subset of another. Ideally, >>> one can use a reasoner (or other tool) that is conforming for a >>> superprofile on the subprofile with no change in the results >>> derived. For profiles such as OWL 2 EL and OWL 2 QL in relation >>> to OWL 2 DL this principle does hold: Every conforming OWL 2 DL >>> reasoner is an OWL 2 EL and OWL 2 QL reasoner (but may differ in >>> performance since the OWL 2 DL reasoner is tuned for a more >>> general set of cases). >> to >>> Note that none of the profiles below is a subset of another. >> but my complaint would have been fixed with By and large, >> different profiles can be distinguished syntactically with there >> being inclusion relations between various profiles. For example, >> OWL 2 DL can be seen as a syntactic fragment of OWL 2 Full and OWL >> 2 QL is a syntactic fragment of OWL 2 DL (and thus of OWL 2 Full). >> None of these profiles below is a subset of another. Ideally, one >> can use a reasoner (or other tool) that is conforming for a >> superprofile on the subprofile with no change in the results >> derived. For profiles OWL 2 EL and OWL 2 QL in relation to OWL 2 >> DL this principle does hold: Every conforming OWL 2 DL reasoner is >> an OWL 2 EL and OWL 2 QL reasoner (but may differ in performance >> since the OWL 2 DL reasoner is tuned for a more general set of >> cases). it was only the one particular sentence that had been >> added that I was asking about >> -JH >> On Aug 5, 2009, at 6:25 AM, Pascal Hitzler wrote: >>> (it evades me, but) done. >>> >>> Diff: http://www.w3.org/2007/OWL/wiki/index.php? >>> title=Primer&diff=25098&oldid=25085 <http://www.w3.org/2007/OWL/ >>> wiki/index.php?title=Primer&diff=25098&oldid=25085> >>> >>> Pascal. >>> >>> Jim Hendler wrote: >>>> Pascal - >>>> You clearly misunderstood me, the sentence you put in the primer >>>> is: >>>> Each of the profiles presented below is a (strict) syntactic >>>> subset of OWL DL, but none of these profiles is a subset of >>>> another. >>>> which is the sentence I am having the problem with! -- the >>>> second part of my response was added to this sentence so as to >>>> clarfiy - so you've made exactly the change I raised my >>>> complaint about... >>>> My first choice would be to do what Ian did in the profiles >>>> document (simply take out the part about syntactic subset and >>>> include the second), >>>> my second choice would be to add a new sentence that fixes the >>>> issue that I am having a problem with (but I agree with Ian >>>> that coming up with something everyone would be happy with would >>>> be too much work and too major a change) >>>> so be good if we could simply go to the change as Ian suggested >>>> thanks >>>> -Jim H. >>>> On Aug 4, 2009, at 5:59 PM, Pascal Hitzler wrote: >>>>> In the primer, the wording is already exactly as the first part >>>>> of Jim's second suggestion. So no further changes to the >>>>> primer at this stage. >>>>> >>>>> Pascal. >>>>> >>>>> >>>>> Ian Horrocks wrote: >>>>>> IMHO this would be a larger and more controversial change than >>>>>> we should be making at this stage. >>>>>> I think that the best solution is the last one suggested by >>>>>> Jim -- to simply say that "none of these profiles is a subset >>>>>> of another". I have updated the document (and response) >>>>>> accordingly. Hopefully Pascal can do the same for the Primer. >>>>>> Ian >>>>>> On 4 Aug 2009, at 16:19, Jie Bao wrote: >>>>>>> On Tue, Aug 4, 2009 at 9:20 AM, Jim >>>>>>> Hendler<hendler@cs.rpi.edu <mailto:hendler@cs.rpi.edu>> wrote: >>>>>>>>>>>>>> Note that each of the profiles is a (strict) >>>>>>>>>>>>>> syntactic subset of OWL >>>>>>>>>>>>>> DL, but none of the profiles is a subset of another >>>>>>>> >>>>>>>> >>>>>>>> While the above is technically correct, I think that some >>>>>>>> people would miss >>>>>>>> the fact that "syntactic" subsets of OWL 2 DL is different >>>>>>>> than the fact >>>>>>>> that you must use the DL restrictions (esp for RL) - so I'd >>>>>>>> suggest one of >>>>>>>> the following three rewordings: >>>>>>>> >>>>>>>> Note that each of the profiles is a (strict) syntactic >>>>>>>> subset of OWL 2's >>>>>>>> syntax, but none of the profiles is a subset of each other >>>>>>>> [[i.e. since >>>>>>>> syntactically OWL DL and OWL Full are same thing, why bring >>>>>>>> up the issue]] >>>>>>>> >>>>>>>> or >>>>>>>> >>>>>>>> Note that each of the profiles is a (strict) syntactic >>>>>>>> subset of OWL 2 DL >>>>>>>> and none of the profiles is a subset of another. We note >>>>>>>> that OWL RL is >>>>>>>> expected to be used primarily with OWL Full semantics, the >>>>>>>> others with OWL >>>>>>>> DL. [[which is clear, but I suspect controversial]] >>>>>>>> >>>>>>> I think the last sentence is important. It would be confusing >>>>>>> if we >>>>>>> just say RL is a syntactic subset of DL, but its reasoning >>>>>>> rules are >>>>>>> in the RDF semantics. It might be bizarre to explain to some >>>>>>> RDF-minded that why in RL we can't say hasBrother >>>>>>> (transitive) and >>>>>>> hasSister (transitive) are disjoint, or hasBrother is >>>>>>> irreflexive, but >>>>>>> we may still apply RL inference rules to an OWL Full ontology >>>>>>> that >>>>>>> says so. >>>>>>> >>>>>>> Jie >>>>>>> >>>>>>>> or >>>>>>>> >>>>>>>> just say >>>>>>>> >>>>>>>> None of these profiles is a subset of another [[and avoid >>>>>>>> the whole issue]] >>>>>>>> >>>>>>>> I'm sorry, but I do consider the quoted line above to a >>>>>>>> problem, and one I >>>>>>>> cannot ignore.... >>>>>>>> -Jim H. >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> On Aug 4, 2009, at 6:42 AM, Ian Horrocks wrote: >>>>>>>> >>>>>>>>> This seems like a good compromise. >>>>>>>>> >>>>>>>>> I have made the relevant changes. The diff is: >>>>>>>>> >>>>>>>>> >>>>>>>>> http://www.w3.org/2007/OWL/wiki/index.php? >>>>>>>>> title=Profiles&diff=25048&oldid=24645 <http://www.w3.org/ >>>>>>>>> 2007/OWL/wiki/index.php?title=Profiles&diff=25048&oldid=24645> >>>>>>>>> >>>>>>>>> Regards, >>>>>>>>> Ian >>>>>>>>> >>>>>>>>> >>>>>>>>> On 4 Aug 2009, at 10:59, Uli Sattler wrote: >>>>>>>>> >>>>>>>>>> >>>>>>>>>> On 3 Aug 2009, at 21:51, Pascal Hitzler wrote: >>>>>>>>>> >>>>>>>>>>> I would really stick to the real/historic explanation (EL >>>>>>>>>>> family). >>>>>>>>>> >>>>>>>>>> we could also add "which is called EL because it is a >>>>>>>>>> *l*anguage (or >>>>>>>>>> *l*logic) that only provides *e*xistential quantification >>>>>>>>>> of variables."? >>>>>>>>>> Cheers, Uli >>>>>>>>>> >>>>>>>>>>> I understand that it's not directly helpful, but at least >>>>>>>>>>> it becomes >>>>>>>>>>> clear that there is some reason to it - and in case >>>>>>>>>>> somebody wants to read >>>>>>>>>>> up on it on the DL literature, he's not lost in the DL >>>>>>>>>>> acronyms ... >>>>>>>>>>> >>>>>>>>>>> In fact I'll add this to the primer as soon as the wiki >>>>>>>>>>> is accessible >>>>>>>>>>> again (it currently seems to be down...) >>>>>>>>>>> >>>>>>>>>>> Pascal. >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> Sandro Hawke wrote: >>>>>>>>>>>>> >>>>>>>>>>>>> I certainly see no problem with adding some minor >>>>>>>>>>>>> explanatory text >>>>>>>>>>>>> along these lines. >>>>>>>>>>>> >>>>>>>>>>>> It seems good to me, too, except for the EL >>>>>>>>>>>> explanation. The reference >>>>>>>>>>>> to EL++ doesn't help anyone. (If you know about EL++, >>>>>>>>>>>> you don't need >>>>>>>>>>>> the explanation; if you don't know about EL++, then >>>>>>>>>>>> knowing the >>>>>>>>>>>> association doesn't help.) >>>>>>>>>>>> So where does the "E" come from? I guess it's from >>>>>>>>>>>> "Existential >>>>>>>>>>>> Restrictions"... That doesn't help very much here. >>>>>>>>>>>> Maybe we can >>>>>>>>>>>> propose a mnemonic? "Extensive", "Efficient", "Easy", >>>>>>>>>>>> "Economical", >>>>>>>>>>>> "Enormous", "Elephantine"... :-) >>>>>>>>>>>> Maybe something like: >>>>>>>>>>>> - The EL profile was orginally named for its use of >>>>>>>>>>>> Existential >>>>>>>>>>>> restrictions, but for a mnemonic, we note that it supports >>>>>>>>>>>> Efficient reasoning, even with Enormous ontologies. >>>>>>>>>>>> ... or something like that. >>>>>>>>>>>> -- Sandro >>>>>>>>>>>>> >>>>>>>>>>>>> From: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk >>>>>>>>>>>>> <mailto:ian.horrocks@comlab.ox.ac.uk>> >>>>>>>>>>>>> Subject: Re: Explain profile acronyms >>>>>>>>>>>>> Date: Mon, 3 Aug 2009 11:40:57 -0500 >>>>>>>>>>>>> >>>>>>>>>>>>>> IMHO this is a not completely unreasonable request. I >>>>>>>>>>>>>> would propose >>>>>>>>>>>>>> to respond by adding to the Introduction of Profiles: >>>>>>>>>>>>>> >>>>>>>>>>>>>> * brief explanations of the acronyms, namely: >>>>>>>>>>>>>> - The EL acronym reflects the profile's basis in the >>>>>>>>>>>>>> EL family of >>>>>>>>>>>>>> description logics [EL++]. >>>>>>>>>>>>>> - The QL acronym reflects the fact that query >>>>>>>>>>>>>> answering in this >>>>>>>>>>>>>> profile can implemented by rewriting queries into a >>>>>>>>>>>>>> standard relational >>>>>>>>>>>>>> Query Language. >>>>>>>>>>>>>> - The RL acronym reflects the fact that reasoning in >>>>>>>>>>>>>> this profile >>>>>>>>>>>>>> can be implemented using a standard Rule Language. >>>>>>>>>>>>>> * the statement "Note that each of the profiles is a >>>>>>>>>>>>>> (strict) >>>>>>>>>>>>>> syntactic subset of OWL DL, but none of the profiles >>>>>>>>>>>>>> is a subset of >>>>>>>>>>>>>> another." >>>>>>>>>>>>>> >>>>>>>>>>>>>> Comments and/or other suggestions? >>>>>>>>>>>>>> >>>>>>>>>>>>>> Ian >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>>> From: Gioele Barabucci <barabucc@cs.unibo.it >>>>>>>>>>>>>>> <mailto:barabucc@cs.unibo.it>> >>>>>>>>>>>>>>> Date: Mon, 20 Jul 2009 14:54:08 +0200 >>>>>>>>>>>>>>> To: public-owl-comments@w3.org <mailto:public-owl- >>>>>>>>>>>>>>> comments@w3.org> >>>>>>>>>>>>>>> Message-ID: <20090720125407.GA32507@cs.unibo.it >>>>>>>>>>>>>>> <mailto:20090720125407.GA32507@cs.unibo.it>> >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Hello, >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> could you please document the meaning of the EL, QL >>>>>>>>>>>>>>> and DL acronyms >>>>>>>>>>>>>>> in >>>>>>>>>>>>>>> the overview section of owl2-profiles and other OWL 2 >>>>>>>>>>>>>>> documents? >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Also, could you explicitly state whether an OWL 2 >>>>>>>>>>>>>>> profile is a >>>>>>>>>>>>>>> strict >>>>>>>>>>>>>>> subset of another? >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Regards, >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> -- >>>>>>>>>>>>>>> Gioele Barabucci <barabucc@cs.unibo.it >>>>>>>>>>>>>>> <mailto:barabucc@cs.unibo.it>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> -- >>>>>>>>>>> PD Dr. Pascal Hitzler >>>>>>>>>>> pascal@pascal-hitzler.de http://www.pascal-hitzler.de >>>>>>>>>>> Semantic Web Textbook: http://www.semantic-web-book.org >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>>> We choose to go to the moon in this decade and do the other >>>>>>>> things, not >>>>>>>> because they are easy, but because they are hard - John F. >>>>>>>> Kennedy, Sept 12, >>>>>>>> 1962 >>>>>>>> >>>>>>>> Prof James Hendler >>>>>>>> http://www.cs.rpi.edu/~hendler >>>>>>>> Tetherless World Constellation Chair & Asst Dean of IT and >>>>>>>> Web Science >>>>>>>> Computer and Cognitive Science Depts >>>>>>>> Rensselaer Polytechnic Institute, Troy NY 12180 >>>>>>>> @jahendler, twitter >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> Jie Bao >>>>>>> http://www.cs.rpi.edu/~baojie >>>>> >>>>> -- >>>>> PD Dr. Pascal Hitzler >>>>> pascal@pascal-hitzler.de http://www.pascal-hitzler.de >>>>> Semantic Web Textbook: http://www.semantic-web-book.org >>>>> >>>>> >>>> We choose to go to the moon in this decade and do the other >>>> things, not because they are easy, but because they are hard - >>>> John F. Kennedy, Sept 12, 1962 >>>> Prof James Hendler http:// >>>> www.cs.rpi.edu/~hendler >>>> Tetherless World Constellation Chair & Asst Dean of IT and Web >>>> Science >>>> Computer and Cognitive Science Depts >>>> Rensselaer Polytechnic Institute, Troy NY 12180 >>>> @jahendler, twitter >>> >>> -- >>> PD Dr. Pascal Hitzler >>> pascal@pascal-hitzler.de http://www.pascal-hitzler.de >>> Semantic Web Textbook: http://www.semantic-web-book.org >>> >> We choose to go to the moon in this decade and do the other >> things, not because they are easy, but because they are hard - >> John F. Kennedy, Sept 12, 1962 >> Prof James Hendler http://www.cs.rpi.edu/~hendler >> Tetherless World Constellation Chair & Asst Dean of IT and Web >> Science >> Computer and Cognitive Science Depts >> Rensselaer Polytechnic Institute, Troy NY 12180 @jahendler, >> twitter > > -- > PD Dr. Pascal Hitzler > pascal@pascal-hitzler.de http://www.pascal-hitzler.de > Semantic Web Textbook: http://www.semantic-web-book.org >
Received on Wednesday, 5 August 2009 14:47:32 UTC