- From: Luc Moreau <L.Moreau@ecs.soton.ac.uk>
- Date: Wed, 14 Dec 2011 12:52:06 +0000
- To: public-prov-wg@w3.org
Hi Simon, Response interleaved. On 12/14/2011 10:46 AM, Simon Miles wrote: > +1 for Proposal 1 > +1 for Proposal 2 > -1 for Proposal 3, as I find "event" a lot clearer than "action" > > Reasons why I think Proposal 3 is unnecessary: > > As far as I understood from earlier discussion, the problem with the > term "event" was that an event is not intuitively instantaneous, > outside of special usage in process algebra etc. I also argued on a > separate thread on Prov-O that "being instantaneous" was something > that could never be honestly asserted, as what is instantaneous for > one asserter may be decomposed by another. We cannot force asserters > to give the time an event occurs to a second if they only know it at > the granularity of a day, and even if they could, many other events > may be separately asserted to happen at different points during that > second. > I think you are mixing imprecision in the measure and the fact that an event is instantaneous. > Proposals 1 and 2 seem to remove the requirement that the events be > instantaneous, so Proposal 3 seems unnecessary. You did say, "It's > also also fine, I believe, to regard this as instantaneous", but I > take this to imply it is also fine to regard an event as > non-instantaneous? I can't see we lose anything by this flexibility. > > I didn't mean this to be optional. I think that proposal 1 and 2 make it clear that it's instantaneous. Satya's alternative suggestion of an 'instantaneous event' is also good for me. (Easier to implement, and closer to the orignal proposal) Luc > Thanks, > Simon > > On 14 December 2011 10:20, Daniel Garijo<dgarijo@delicias.dia.fi.upm.es> wrote: > >> +1 for proposals 1 and 2. >> +1 for proposal 3. As long as we agree on what we are dealing with, I don't >> have anything against using another name. >> And if we can avoid confusion with other ontologies, then even better. >> >> Best, >> Daniel >> >> >> 2011/12/10 Satya Sahoo<satya.sahoo@case.edu> >> >>> +1 for proposal 1 >>> +1 for proposal 2 >>> -1 for proposal 3 (I believe explicitly stating instantaneousEvent is >>> better option than re-labeling as action. Action, similar to event, has >>> multiple connotations including instantaneous and long running) >>> >>> Best, >>> Satya >>> >>> >>> On Sat, Dec 10, 2011 at 12:50 AM, Reza B'Far (Oracle) >>> <reza.bfar@oracle.com> wrote: >>> >>>> +1 on Proposal 1. I think this actually resolves 2 other issues I >>>> brought up a couple of months ago (but let go) -- providing an "origination >>>> time" which is important in legal/financial applications as well as making a >>>> distinction between the very first time something is created vs. the >>>> subsequent times (for example, ratification of a law). Anyways, great >>>> proposal. I realize the nuance in the comment is a bit different than what >>>> I'm saying, but it doesn't matter. There is now some mechanism to solve >>>> what I need :) >>>> >>>> +0 on Proposal 2. Clear semantics and good proposal to understand, but I >>>> think the verbiage is awkward for beginners. I could change to +1 if the >>>> verbiage is made easier. >>>> >>>> +1 on Proposal 3 conditional on going through Prov-DM and assuring that >>>> we don't use the words "actions" or "action" in any other context in the >>>> examples, etc. [hope that makes sense] >>>> >>>> >>>> >>>> On 12/9/11 10:36 AM, Luc Moreau wrote: >>>> >>>> Dear all, >>>> >>>> Several of you, including Satya, Tim and Jim have raised various >>>> concerns about events. Here are some proposals to tackle these >>>> concerns in prov-dm WD3. Can you express your support in the usual >>>> way, we will confirm the outcome at the next teleconference. >>>> >>>> CONTEXT: >>>> >>>> 1. The concept of event as defined in prov-dm is instantaneous. On the >>>> other hand, other ontologies define the concept of event, e.g. LODE >>>> [1], as not instantaneous. This causes confusion, and risks >>>> hampering adoption. >>>> >>>> 2. The prov-dm document (WD2) is defining 'generation' in a >>>> conflicting manner. On the one hand, in [2], it states that >>>> generation is an event, so is instantaneous. On the other hand, in >>>> [3], it states that completion of generation is the event. >>>> >>>> In a generation, what we care about is the point at which the entity >>>> becomes available for consumption by others. Before that, it's not an >>>> entity yet (or it is not this entity being generated). So, calling >>>> the whole of generation an event (forget the choice of word for now) >>>> is not what was intended. The event is the point at which generation >>>> is complete. >>>> >>>> This is actually nice reflected in Olaf and Jun's provenance >>>> vocabulary [4], where they have a similar concept, called Data >>>> Creation defined as: >>>> >>>> DataCreation is a class that represents the completed creation of a >>>> data item. >>>> >>>> >>>> Note the choice of word *completed*. >>>> >>>> PROPOSALS: >>>> >>>> >>>> We therefore propose to change the definition of Generation [2] as >>>> follows. >>>> - With proposals 1 and 2, resolve the conflicting definitions around >>>> generation (and use) in prov-dm. >>>> - With proposal 3, adopt another name for event. >>>> >>>> >>>> >>>> PROPOSAL 1. Adopt the following Definition for generation. >>>> In PROV-DM, a generation record is a representation of a >>>> world event, the completed creation of a new entity by an >>>> activity. This entity did not exist before this event; this entity is >>>> available for usage after this event. >>>> >>>> >>>> Comment: With this, we are not saying that creation of an entity is >>>> the event, it's the completed creation that is an event. It's also >>>> also fine, I believe, to regard this as instantaneous. Also, if >>>> somebody wants to model the actual creation, it is also fine, they can >>>> use activities for that. >>>> >>>> For usage, we would take a similar approach. In the provenance >>>> vocabulary, they use the completed access to a data structure, but >>>> this is not right for what we want. Instead: >>>> >>>> PROPOSAL 2. Adopt the following Definition for usage. >>>> In PROV-DM, a usage record is a representation of a world >>>> event: the start of an entity consumption by an activity. Before this >>>> event, this entity was not consumed or used in any form or shape by >>>> the activity, totally or partially. >>>> >>>> Comment: These definitions are now exactly in line with those in [3]. >>>> >>>> >>>> PROPOSAL 3. Replace the word event by action. >>>> >>>> Comment: So, prov-dm would define four actions: entity >>>> generation/entity usage/activity start/activity end, which are all >>>> instantaneous. These actions would have "effects" on the system in >>>> the sense that they change the entities and activities it contains. >>>> >>>> Assuming proposal 3 is adopted, obviously, the text of proposals 1 and >>>> 2 would use the word 'action'. >>>> >>>> >>>> Can you express your support, or counter-proposals, by Wednesday midnight >>>> GMT. >>>> Assuming there is support, we would incorporate all these changes before >>>> XMas. >>>> >>>> Best regards, >>>> >>>> >>>> Luc >>>> >>>> [1] http://linkedevents.org/ontology/ >>>> [2] >>>> http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#dfn-Generation >>>> [3] >>>> http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#types-of-events >>>> [4] >>>> http://trdf.sourceforge.net/provenance/ns.html#sec-DataCreationClasses >>>> >>>> >>> >> > > > -- 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 Wednesday, 14 December 2011 12:55:08 UTC