- From: Luc Moreau <L.Moreau@ecs.soton.ac.uk>
- Date: Thu, 22 Dec 2011 16:40:34 +0000
- To: public-prov-wg@w3.org
Hi James, Thanks for drafting the semantics, it will help us pin down issues in the data model. I have a few general questions first, about the alignment of the semantics and the data model. As agreed in the call, I am sharing them by email. 1. PROV-DM identifies three levels 1. things in the world 2. entities, which are characterized things 3. entity records, which are the records we create as part of a provenance record It seems that things are not mentioned in your document. I believe they are important to define specialization/alternates 2. PROV-DM tends to talk about events, whereas your semantics focuses on time. PROV-DM assumes the existence of a mapping from events to time. Is it possible to align both? 3. Used or Generated don't seem to have activities. It seems counter-intuitive. Can you clarify? Cheers, Luc On 12/19/2011 06:32 PM, James Cheney wrote: > Hi, > > I've revised the formal semantics draft at: > > http://www.w3.org/2011/prov/wiki/FormalSemanticsStrawman > > This is still work in progress, and there are a number of areas flagged TODO or left out of scope. However, I think it would be good to get feedback on this before filling in all of these. > > The main issue (which is still in flux in the discussions on the mailing list) is the treatment of wasComplementOf/viewOf/foobar. I've kept the old section that attempted to formalize wasComplementOf and I've added two new sections that discuss viewOf and foobar relations. > > The current treatment suggest that there is a mismatch between my intuition of entities as things that change over time, and entity records that describe temporary states of entities, vs. the view taken in PROV-DM where wasComplementOf and company are assertions about how different entity records are related. > > I think that handling these relations (as currently described in PROV-DM) will require adding something to the formal semantics, along the lines of a relation linking different entity records that are "views" of the same underlying (changing) thing. > > I also think that it might be best to simplify interpretations to be time-independent, which would align with viewing them as URIs. > > Comments are welcome. > > --James > -- Professor Luc Moreau Electronics and Computer Science tel: +44 23 8059 4487 University of Southampton fax: +44 23 8059 2865 Southampton SO17 1BJ email: l.moreau@ecs.soton.ac.uk United Kingdom http://www.ecs.soton.ac.uk/~lavm
Received on Thursday, 22 December 2011 16:41:15 UTC