[Crm-sig] ISSUE: "P43 has dimension" should apply to E1 Entity, not E70 Thing

Vladimir Alexiev vladimir.alexiev at ontotext.com
Wed Nov 23 12:21:36 EET 2011


Hi Stephen, thanks for the feedback!

> is not the event that has the dimension but the object: It has a value
measured
> as part of the sale event.

Imagine this situation: "A cow and some chicken were sold for $5000".
- What if I don't know the complete set of objects that was sold? (I can
identify the cow, but don't even know the number of chicken)
- What if I don't know the split of the price amongst the objects?
- Do I need to make a collection of these objects, only so I can say the
price? Then why is "P24 transferred title of" multi-valued?

> The number of people in a group should always be calculated at query-time
> otherwise we end up with monotonicity problems when we combine data
sources.

What if you don't know the people's identities? What if you don't even know
the exact number?
E.g. "approximately 5000 people were evacuated as a result of a dangerous
goods incident"

This reminds of a joke: a Geography professor really wanted to fail a
certain student.
- Prof: What's the population of China?
- Student: Over 1.3B. It was 1330044605 as of mid-2008, and current best
projections put it at 1553235624 people +-1%
- Prof: well, but can you name them by name?

> - from project management: cost, effort. [These are dimensions of the
Thing
> the project produces.]

There are projects that produce nothing (failed ones).
And the correlation between the effort/cost spent on something and its "true
cost" is not straightforward.
That's the point of Earned Value Management: to be able to track planned vs
actual vs earned cost.

> - from risk management: probability, impact. 
> [the idea of the thing happening,
> not an Event that has not happened (and indeed may never happen!).]

So a Risk would be maybe E90 Symbolic Object that, if materialized, begets a
Risk that is E5 Event?
You have a point here... Though maybe CRM should also be able to model
*hypothetical* events.

But still: upon materialization, the conceptual risk's Impact will become
the risk event's Loss.
How do you model this Loss without going into details on what exactly was
lost (and I may not want to go to such details)

> - from accidents and dangerous goods: number of deaths [use death event
and
> count], number injured [this is an activity involving the injured- then
> count them], area evacuated [Site], number of people evacuated [number of
> participants in the activity], quantity (and type[E55 Type!]) of good
> spilled [dimension of the Thing spilt], etc.

You may be able to "individuate" some of these into Things, but in many
cases you cannot.
People want to talk about "approx number of injured/dead" even before they
find all the bodies and write off the unfound as "Missing persons".

I think there are also a lot of examples from Physics and Chemistry that
need to measure Events, not the (often unknown or uncountable) things the
event operates on.
E.g. intensity of chemical reaction, acceleration (rate of change of the
speed of a movement), rate of Absorbed Dose of radiation, etc.

Look at http://qudt.org/ in section "The SI System". Search for "per
second". I think these units measure Events, not Things.

> In general temporal entities can only be measured in time and all the
> properties for locating and measuring them in time are already in the CRM.

Do you mean only Duration and dates (inner and outer bounds)?
How do you prove the statement "can only"?

And finally, how do you explain this: "P39 measured" points to E1, "P43 has
dimension" is a shortcut thereof, but P43 points lower than E1.
http://personal.sirma.bg/vladimir/crm-graphical/#measurement 



More information about the Crm-sig mailing list