Re: specializationOf and inference

> On Jul 28, 2017, at 5:06 AM, <Simon.Cox@csiro.au> <Simon.Cox@csiro.au> wrote:
> 
>> In Prov, prov:Entities stand for aspects/views of things, not the things themselves.  
> 
> Is that always the case, or only when there are alternateOf of specializationOf relationships involved? 
> In prov-o the definition of Entity is 
> 
> "An entity is a physical, digital, conceptual, or other kind of thing with some fixed aspects; entities may be real or imaginary." 
> 
> This appears to say that Entity can be the thing itself. 


Hi,

That's correct, the definition in prov-o and prov-dm does say that.  We had a lot of discussions about this since (as this thread shows) one can run into lots of problems if representing information about "the same thing" at different times, from different points of view, at different levels of granuality (e.g. <#me> vs. <#my-left_ear>), etc.  

I think where we landed was providing the simple view (an entity is a thing with some fixed attributes) rather than something like "an entity is an aspect of / represents some thing with some fixed attributes".  I don't remember the decision process, but in general I think the decision was to present something simple in intuitive terms whenever possible, rather than scare everyone off first thing.  

Of course, when we are talking about a representation language, a prov:Entity URI is not the thing itself; just a representation of it.  Where things get slightly tricky is that the alternate and specialization relations are really relating the representations, not the underlying things.

--James