- From: Christine Golbreich <cgolbrei@gmail.com>
- Date: Mon, 2 Mar 2009 16:36:47 +0100
- To: Michel Dumontier <michel.dumontier@gmail.com>
- Cc: W3C OWL Working Group <public-owl-wg@w3.org>
Hi Michel, Since I did not get any feedback from the group providing another real UC with an example of local reflexivity, I propose to keep UC#3. Although the terminology in paper [1] might not be really explicit or universal, I suggest to stick on it to fix the example. Thus, I have replaced RingMolecule by RingAtom see the diff http://www.w3.org/2007/OWL/wiki/index.php?title=New_Features_and_Rationale&diff=18595&oldid=18593. There are 2 advantages: 1) This matches the semantics of a RingAtom accordiing to the SWRL rule proposed to identify a RingAtom in this paper. 2) This illustrates the local reflexivity feature by a simple example from a real referenced UC (BTW. the purpose here is not to recognize a ring atom, or to discuss global restrictions, but only to give an illustrative example ) If you don't like this example, since a Cyclic Structure is asserted to be Local reflexive in Table 1, if you prefer, we may then replace RingMolecule by RingStructure ... like in that table. Please, let me know how you feel about it. Option 2. move UC#3 from the 2.2.1 F4: Self Restriction for illustrating other features in another section BUT provide another UC for Local refelexivity Option 3. remove UC#3 from the document Christine 2009/1/23 Michel Dumontier <michel.dumontier@gmail.com>: > Hi all, > > In reference to [1], comment [2] describes the details as erroneous. While > the comment is valid, what is more troubling is that the conclusion drawn is > incorrect. The cited references [3][4] specifically discusses how the > proposed set of OWL (1.1) features was insufficient to capture the ring > structure between connected atoms, and must be dealt with either with SWRL > rules, and possibly with description graphs. So, while reflective properties > maybe useful in other domains (for instance, to assert that the mereological > relation has_part is reflective), it is not particularly useful, at this > time, in the domain of chemical structure. > -=Michel=- > [1] http://www.w3.org/TR/owl2-new-features/#Use_Case_.233_-_Classification_of_chemical_compounds_.5BHCLS.5D > [2] http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0021.html > [3] http://sunsite.informatik.rwth-aachen.de/Publications/CEUR-WS/Vol-258/paper28.pd > [4] http://www.webont.org/owled/2008dc/papers/owled2008dc_paper_20.pdf >> >> * Local reflexivity (can now define a chemical ring, although I defer to >> chemist opinions on this such as Egon (already commented), Colin?, Michel?) > > > > is erroneous. In our paper, we described how OWL2 was not sufficient to > fully describe molecular structure because it was not possible to specify a > partial ordered path > >> >> * Qualified Cardinality (histone example: a H3K4m3 has been methylated >> exactly 3 times) >> * The new reflexive, irreflexive, and asymmetric property axioms >> * Features that increase compatibility of OWL 2 with OBO and SNOMED! >> >> Thank you for your excellent work! >> >> HCLS IG has a few applications where rules are used in combination with >> OWL/RDF. In general, being able to build OWL out of RIF is an appealing form >> of interoperability. So, I have some concerns about the fact that in RIF the >> xsd numeric types have disjoint value spaces (as in XSD1.1, unlike current >> OWL 2 drafts). I am also concerned to learn that there are different data >> types in RIF versus OWL. For example, maybe OWL-RL could be implemented on >> top of RIF, but this could become impractical if there are data type issues. >> I hope that the data type issues between OWL and RIF can eventually be >> resolved. >> >> BTW, it would be useful to have a short overview of the document set that >> briefly explains the content/purpose of each document. >> >> Best, >> Scott >> >> Typo in http://www.w3.org/TR/owl2-new-features/ : >> E.g.; a frontal lobe is part of a brain memisphere or a car is part of a >> car memisphere -> hemisphere >> >> P.S. Handy for HCLS folks!: >> Property chain inclusion (from RequirementsDraft): >> SubPropertyOf( PropertyChain( locatedIn partOf ) locatedIn ) (UC#7) >> >> If x is locatedIn y, and y is partOf z, then x is locatedIn z; for example >> a disease located in a part is located in the whole. >> >> -- >> M. Scott Marshall >> http://staff.science.uva.nl/~marshall >> http://adaptivedisclosure.org >> >> > > > > -- > Michel Dumontier > Assistant Professor of Bioinformatics > http://dumontierlab.com > -- Christine
Received on Monday, 2 March 2009 15:37:25 UTC