PG
So, I'm trying to extend our use of ARIS from process modelling into deeper information modelling. We have used variants of the IE Data model and the eERM attribute allocation diagram to describe various aspects of our information landscape.
When it comes to different representations of the same data, I'm running into a wall. What I'd like to do is to model
- what ERM attributes (belonging to one Entity Type) are similar to (for the business level) other ERM attributes;
- what business level clusters/entities/attributes are represented by logical or physical level entities/attributes (where we move from business terminology to system table and column names);
- what physical level attributes in one system are copied into which physical level attributes in another system (or even the same, when tables represent overlapping concepts).
The wall I am running into is that I can find almost no connection types that are actually permitted to connect two ERM Attribute objects. I would be happy enough to manage and show connections as a Matrix model, but that requires some connection type with a reasonably relevant name.
The only option I find is the "propagates" relation in either the Attribute allocation diagram (Solution design) or the OMT object model: that's a reasonable fit to case 3 above, but a less felicitious name for cases 1 and 2. Are there any other options apart from those that use model types that connect ERM attributes to some arbitrary intermediate objects? If so, how do I generate a Matrix model from those models, for a more compact representation?
Hi Pontus Gagge
what business level clusters/entities/attributes are represented by logical or physical level entities/attributes (where we move from business terminology to system table and column names)
Here is how I plan to do this, this is a work in progress.
I'm linking Cluster data/model and entity types in IE Data model
"cluster data/model" are mapped to "data objects" in BPMN Models or "business objets" in Archimate models : so they are our primary source of "business objects" in our models. (image 1 below)
If we need to further describe "business objects" for example by adding attributes, we create an "entity type" object type, associate it to the corresponding "cluster/data model" in an IE data model (image 2 below)
.PNG)
Adding attributes in another IE Data Model which is "application" indepedent (think of a conceptual data model)
If we need to model application specific data model, we use another Aris model : a "data type diagram (solution design)". This is not a physical model (Aris has other models for this), it's more to record what data is in which application , and how the application names that data.
Linking application-independent conceptual data model and application-dependent data model is done through a matrix model :
.PNG)
We're just at the beginning of our journey and I'm not sure if this approach is sustainable in the long run because of the number of differents models and objetcts to be maintained (and note that I'm not even using Aris tables models to describe data at a more physical level)
Any comments are appreciated ;-)
Hi Michel,
very well described. One suggestion: Instead of adding the attributes in the IE data model you could hide them in an assigned "ERM attribute allocation diagram". This way you can concentrate on the relationships between the entity types in the IE data model. This is pretty much the same pattern as between the EPC and FAD.
One might consider, if representing "K attributes" on the IE data model level is desired or if you show them simply as "D attributes". I definitely agree that you do not show "FK attributes" on that level. It would be redundant to the relationships between the entity types.
Thanks M. Zschuckelt
Pontus Gagge you can read here discussions on modeling data flows and transformations :
