RE: [Crm-sig] P53 and P56; shortcuts; propertyChainAxiom

[Vladimir Alexiev]

> About E18: What would be an example of an E18 Physical Thing that
> doesn't belong to any subclass of E18, which are E19 Physical Object,
> E24 Physical Man-Made Thing and E26 Physical Feature?

There doesn't need to be such example: these 3 classes could provide a (disjoint) covering of E18.

> Are there things that are E19 and E26 at the same time?

The scope note of E18 implies that no: "Most instances of E19 Physical Object can be moved (if not too heavy), whereas features are integral to the surrounding matter."
However, CRM doesn’t define any disjointness, and Martin has argued that ECRM over-commits by declaring owl:disjointWith, and I sometimes find it useful to use various seemingly disjoint classes together (e.g. equate a Collection with the Legal Body holding that collection).

> P53 "has former or current location" has the domain E18, but the fully
> developed path starts with E19. is it because going though E9 Move
> wouldn't make sense?

Yes: Moving a feature doesn't make sense: you could move the containing object, but the feature is assumed to stay fixed on that object.

(It was explained that it is for that reason P55 has current location and P54 has current permanent location don't apply to features.
Which is quite unexpected, and I argued should be reflected in their scope notes)

> Take an E18 that is not an E19, for
> example the cave “Ideon Andron” in Crete (E26). Is it correct that in
> this case P53 is not a shortcut of "E19 P25i moved by E9 Move P26 moved
> to / P27 moved from E53"?

CRM shortcuts are the same as owl:propertyChainAxiom: if the longcut holds, that implies the shortcut. However, the shortcut could also exist without any longcut.
So your statement is not quite right, it should be: "in this case P53 is not a shortcut of any longcut".

CRM does not declare any shortcuts as owl:propertyChainAxiom.

Even ECRM/EFRBROO (which declare inverses, transitives and disjoints), don't declare any owl:propertyChainAxiom
- E.g. in a previous EFRBROO version, R26 mistakenly was declared a subprop of P108 and P2
http://erlangen-crm.org/docs/efrbroo/120730/objectproperties/R26producedthingsofty___2094266564.html
- This was corrected by Justyna Walkowska, but now there's no connection between these props:
http://erlangen-crm.org/docs/efrbroo/latest/objectproperties/R26producedthingsofty___366730490.html

> Is a tsunami wave a (moving) feature of an ocean, or not?

A very good question. Here are some more:
- if a Wave is a feature, how about the waterdrops floating above the wave?
- is the water falling from a waterfall a feature of the waterfall, or a separate thing?
- how about after the water flows down the river? Does it now have any relation to the waterfall?

Luckily CRM is not intended to model such dynamic situations:
"The CRM is generally not concerned with amounts of matter in fluid or gaseous states."

> Is P53 the only shortcut that can be extended only for a subset of its domain or range?

That's not the only case. I've started an analysis of domain/range "discrepancies" between shortcut/longcut, but haven't completed it.
There are 20 CRM shortcuts (not counting subclasses of E13), and for about half of them the domain & range are not exactly the same.

Below you describe the other situation P53 is inovolved in (in this case in the longcut):
longcut: E18-P59-E53-P53i-E18, shortcut: E19-P56-E26

> Is the following correct? "E19 P56 bears feature E26" can
> always be extended to "E18 P59 has section E53 Place P53i is former or
> current location of E18"

No. Typically the longcut bears more info than the shortcut, so there's no point in such extending.
You only need to use this longcut if you have some extra info about the E53 Place, e.g.
- a Place Appellation such as id or name
- the fact that the same place is the location of several features

The domain & range of the longcut above means that one can define sections not only on E19 Physical Object but on any E18 Physical Thing, including E26 Feature.
The domain & range of P56 as defined restrict the shortcut to only a subset of the subjects & objects to which the longcut can be applied.
Whether this is reasonable or not is a question for investigation.

> Taken together, "E19 P56 bears feature E26" can be
> extended to "E19 P58 has section definition E46 Section Definition P87i
> identifies E53 Place P53i is former or current location of E26".

Yes, here you're chaining two longcuts (P59 is a shortcut but also participates in a longcut).
No, shortcuts are not extended to longcuts. Rather, longcuts (should) imply shortcuts (e.g. through propertyChainAxiom)

[Martin Scholz]

Hi Vladimir,

2013/3/11 Vladimir Alexiev <vlad...@sirma.bg>:

> That's not the only case. I've started an analysis of domain/range "discrepancies" between shortcut/longcut, but haven't completed it.
> There are 20 CRM shortcuts (not counting subclasses of E13), and for about half of them the domain & range are not exactly the same.

I am really looking forward to your analysis!

Some time ago, when we where debating about releasing an OWL 2 version
of the ECRM, Mark posted an (incomplete) list of shortcuts:
<https://groups.google.com/d/msg/erlangen-crm/ojF1l8EhoPc/2bzmgIbW3sQJ>.
Maybe it is of any help to you.

This post also mentions some shortcuts not explicitly stated in the CRM spec.
I am wondering about the inheritence of shortcuts and if there are any
relevant entailsment. The post mentions shortcuts having
subproperties. While it sounds logical to me, that the subproperties
must be shortcuts as well, you can not infer/associate the subproperty
from/with the longcut (although the longcut may represent exactly
that).

We are still planning to release a OWL2 ECRM. So any findings will be
appreciated! :)

Regards
Martin

[Vladimir Alexiev]

> Mark posted an (incomplete) list of shortcuts:
> <https://groups.google.com/d/msg/erlangen-crm/ojF1l8EhoPc/2bzmgIbW3sQJ>.

Thanks! I didn't know about this list.
If I can make some time to work on this, I'll compare to my list...

At the end Marc notices a domain discrepancy ("The end of this more elaborated path is E42 and not E41...").

Here's another one:
2) "P43 has dimension" is a shortcut for E70 -> P39 -> E16 Measurement-> P40 -> E54
The domain of P39_measured is P1, which doesn't match the domain of P43_has_dimension, which is E70.

So you can measure all kinds of entities, but you can record the dimensions only of E70 Thing! E.g.
- you cannot record the speed of an E7 Activity "Driving a car"
- you cannot record how many fatalities the E5 Event "Eruption of Vesuvius" resulted in
- you cannot record how many members an E74 Group has

I wrote about this half a year ago (ISSUE: "P43 has dimension" should apply to E1 Entity, not E70 Thing),
which was dismissed through all kinds of philosophical arguments by Steve Stead (IMHO unconvincing).

> I am wondering about the inheritence of shortcuts and if there are any
> relevant entailsment. The post mentions shortcuts having
> subproperties. While it sounds logical to me, that the subproperties
> must be shortcuts as well

Not necessarily:
- Both a longcut and a subproperty provide more info than a shortcut
- So both a propertyChainAxiom and a subproperty can infer the shortcut, but not the other way around.
Now, if you have a *superproperty* of the shortcut, the superproperty can be inferred from the longcut by chaining the two inferences.

This said, in many cases a subproperty CAN participate in a longcut.

Let's see an interesting example:
The scope note of "P1 is identified by" says "... fully developed (i.e. indirect) path through E15 Identifier Assignment" but doesn't say which properties participate in the longcut.
Look at these two graphic situations:
- http://personal.sirma.bg/vladimir/crm-graphical/#apellation
- http://personal.sirma.bg/vladimir/crm-graphical/#object_number
It seems clear the longcut should be
E1 - P140i_was_attributed_by - E15 Identifier Assignment - *P37_assigned* - E42,
and should NOT involve P38_deassigned.

However, P38 is a subprop of P141_assigned, just like P37 is ;-)