- From: Alan Ruttenberg <alanruttenberg@gmail.com>
- Date: Mon, 19 Mar 2007 16:31:12 -0400
- To: John Barkley <jbarkley@nist.gov>, public-semweb-lifesci@w3.org
Just had a discussion with John, summary: Minor, stylistic: Class names are often written as singular e.g. Cells -> Cell. Single URI for each class, based on the id. The name of the thing goes to the label, and an annotation property records the abbreviation. part_has_cell, cell_has_molecule_within, part_has_molecule_in_region, become subproperties of ro:location_of (from relation ontology) We had a discussion of the allValuesFrom restrictions. The first was re: his previous mail. He was using separate allValuesFrom which were being interpreted as a conjunction. Instead he would change these to a single allValueFrom of the union. The second question was whether the allValuesFrom was appropriate. It seemed appropriate in some cases, not in others. For the part/part relations, we assumed that the atlas was exhaustive, so the restriction makes sense. We're not as sure about source/target, part/cell, cell/molecule relations. These are probably not exhaustive, so someValueFrom alone would seem to make more sense. He will ask Mikhail and adjust the model accordingly. The "by_" classes will be removed. -Alan On Mar 19, 2007, at 10:15 AM, Alan Ruttenberg wrote: > > Hi John, > > Great progress! > > I think I would choose a single name and then have e.g, id and > abbreviation be properties of it. You can use annotation properties > for these. Then remove the "by" layer of the ontology - these lists > can be regenerate on demand via e.g. a sparql query. > > I notice that you use has_part and part_of in the definitions. > Remember that each is a all-some statement, so perhaps verify that > that is true. e.g. a has_part some b = every a has some b as a > part. a part_of some b = every a is part of some b. Sometimes only > one or the other direction is true. > > Speak to you soon, > > Best, > Alan > > > On Mar 19, 2007, at 8:18 AM, John Barkley wrote: > >> >> For another one of my action items, I put on the wiki demo page: >> http://esw.w3.org/topic/HCLS/HCLSIG_DemoHomePage_HCLSIG_Demo >> a reference to the initial bams class style model and an updated >> bams instance >> style model. In both, I made the properties "has_source" and >> "has_target" >> transitive. Please let me know if this is not correct. >> >> There are no instances in the class style model. It is in the >> style of the >> Pizza tutorial >> (http://www.co-ode.org/resources/tutorials/ProtegeOWLTutorial.pdf). >> You may notice that the model is OWL Full, but that's only because >> there is an >> annotation property "has_bams_URL". Reasoners generally ignore >> annotation >> properties. >> >> jb >> >> >> > >
Received on Monday, 19 March 2007 20:31:59 UTC