Deriving opposite structural relationships ?

[Alberto D Mendoza]

Hi folks, 

I'm missing something basic here,  derivations for Structural Relationships are ordered by strength with the weakest being the derived relationship. The general rule being:

If two relationships p(a,b):S and q(b,c):T exist, with S and T being structural relationships, then a relationship r(a,c):U can be derived, with U being the weakest of S and T.

Unfortunately, the rule does not indicate whether direction matters, and what is the rule to follow if this is the case.  There are derivations rules when both structural and dependency relations with opposite directions meet, but not for structural only. 

What am I missing?

[Mastering ArchiMate]

Direction is essential

p(a,b) and q(b,c) are ways to describe direction. Opposite direction would be p(a,b) and q(c,b)

--
You received this message because you are subscribed to the Google Groups "ArchiMate" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-archimate-f...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/open-archimate-forum/85bc6bbc-2543-4885-ba25-c3460484699an%40googlegroups.com.

[ADM@GMAIL]

I think I found my answer since I was using Actor assignment so I could only derive one way, but for the case where 

p(a,b):S and q(c,b):T and with S and T being structural relationships, how do you decide the derived relationship direction? r(a,c):U or r(c,a):U. 

I’m guessing the weakest link sets directionality?  What if both relationships are the same type?

Alberto

On Oct 8, 2020, at 4:33 AM, Mastering ArchiMate <mastering...@gmail.com> wrote:

Direction is essential

[Kalin Maldjanski]

You cannot derive r(a,c):U nor r(c,a):U from that actually, unless you set your own rules.

i.e. you have B is part of A and B is part of C which is perfectly possible, but you cannot infer anything from that about relationship between A and C. You can use the semantics of the relationships and the elements and create more specific custom inference rules.

Btw on the comment "I was using Actor assignment so I could only derive one way," current derivation rules does not take into account Element types, thus being an Actor or not does not really matter.

Also note that now in ArchiMate 3.1 all relationships are directed except Association Relationship which can also be explicitly marked as directed.

[Mastering ArchiMate]

On 8 Oct 2020, at 23:34, ADM@GMAIL <alberto....@gmail.com> wrote:

I think I found my answer since I was using Actor assignment so I could only derive one way, but for the case where 

The element type does not matter.

p(a,b):S and q(c,b):T and with S and T being structural relationships, how do you decide the derived relationship direction?

p(a,b):S means: there is a relation S from a to b

q(c,b):T means: there is a relatiion T from c to b

There is no derivation from a to c or vice versa, because the direction of S is opposite to that of T. 

a ->(S)-> b <-(T)<- c

The directionality is inherent to the relation (in this case S or T) and it defines if a derivation is possible.

r(a,c):U or r(c,a):U. 

Sp, no such ‘choice'

I’m guessing the weakest link sets directionality?  

No. The directionality is ‘input’ for the derivation mechanism, not ‘result’.

What if both relationships are the same type?

Doesn’t matter.

G