- From: Alan Ruttenberg <alanruttenberg@gmail.com>
- Date: Thu, 5 Mar 2009 10:28:38 -0500
- To: Bijan Parsia <bparsia@cs.manchester.ac.uk>
- Cc: Michael Schneider <schneid@fzi.de>, "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, public-owl-wg@w3.org
On Thu, Mar 5, 2009 at 9:53 AM, Bijan Parsia <bparsia@cs.manchester.ac.uk> wrote: > On 5 Mar 2009, at 14:22, Alan Ruttenberg wrote: > >> On Thu, Mar 5, 2009 at 7:59 PM, Michael Schneider <schneid@fzi.de> wrote: > > [snip] >> >> I can see her having a question about this. For instance the top >> google hit for database keys >> http://databases.about.com/cs/administration/g/key.htm says: >> >>> Definition: A database key is a attribute utilized to sort and/or >>> identify data in some manner. Each table has a primary key which uniquely >>> identifies records. Foriegn keys are utilized to cross-reference data >>> between relational tables. >> >> If the key is not functional then the sort order may not be well defined. > > Er... use to sort *or* identify. > > Multiple key values can be used to identify in several different ways. For > example, if one is doing entity reconciliation. We don't *have* tables, so > keys identify *entities* not records. > > Similarly, with sort. There is no notion of sorting in OWL, but there are > many ways to handle multiple sort keys for an entities (e.g., lowest wins, > the entity appears in each position, etc.). These are all well defined. And > then of course one could impose functionality. > > Again, since OWL is used to model multiple databases in various states and > to integrate them, it's important to allow these situations to be > represented. I completely agree about the need to let these be represented. The question that Marijke raises isn't about that but about how we document the feature. -Alan > >> I think she would have a better argument about our nomenclature being >> confusing if she could cite authoritative sources that define key >> properties as functional. Unless someone objects I will follow up with >> her along these lines off list. > > I think she has agreed that it's a judgement call and our judgements > disagree. Thus, I think she has accepted our response. I am very reluctant > to have the group debate her on this point. > > (If you want to pursue it personally...fine.) > > If there's doubt whether she "agrees to disagree" then that's worth a > (group) follow up. > > Cheers, > Bijan. >
Received on Thursday, 5 March 2009 15:40:39 UTC