doremus-ontology
doremus-ontology copied to clipboard
DOREMUS-ANR
Can we clarify things around shortcuts and simple representations ?
@rtroncy you recently said to the nice folk at Erlangen that "we love shortcuts". Can we clarify when shortcuts are useful and how to model and implement them ?
A simple example is u24_has_award. To state that a CD received a Diapason d'Or, we model it through a long path with M34_Award_Assignment, see schema.
But you can see in this schema that we also created u24_has_award : F24_Publication_Expression u24_has_award M36_Award "diapason d'or"
The "long path" gives us more info, e.g. the author of the award ("Diapason"), the year, place.
The shortcut just gives us a handy way to answer the simple question "what awards did this CD receive" and also "which CDs received a Diapason". If we want to add a year in our request, we need the long path, so u24 is not a real shortcut, it's a simpler representation, hence the broad title of this issue.
(Yet, I use the word "shortcut" because on a similar case, R40_has_representative_expression, the FRBRoo documentation speaks about shortcut for F1 - R50i - F42 - R41 - F2)
So :
- what is the right definition of a shortcut ?
- how do we implement it ?
- is it ok to create simple representations that kind of double the long path ?
@rtroncy here is an example of a simpler representation together with the long path.
This schema shows the Derivation Type Assignment we decided together on page 1. On page 2, I added a "simpler" property : Uxx_is_derivation_type_of. We may not want to ask for "all the possible derivation types of a Work", since it can only have one per institution. OK, institutions may have different points of view, and this query may highlight this, but this is a very limited option. Therefore, the property is oriented the other way around : "what works are (transcription, reduction, etc.)".
So my question is two-fold :
- is this property useful ?
- is its implementation correct ? (it is not implemented as a real shortcut but as a simpler property - simpler in the sense that we lose information : author, date, place)
@rtroncy I come back to you regarding "shortcuts". Since we may have a number of shortcuts it would be nice to have a clear view on this topic. For me there are 2 ways we could consider a shortcut, either as a chain of properties or as a distinct, independent property.
Therefore I have 2 distinct questions :
- Is it ok to have both types of shortcuts? does it make sense? is it useful? Or should we do with one type only, i.e. chained properties ?
- how do we implement chained properties? how do we create them in Protégé?
Thx
As I briefly replied to you in Montpellier, this is not for now and you should not be too much concerned about it. If you need a short direct property, just create it. We will always be able to strengthen the definition of a property by adding equivalence axioms composed of property chains. We should aim to release asap the version 1 of the DOREMUS ontology and such axioms can be part of future releases. I have assigned myself this action and I will keep it open, working on it after the release of the first version of the ontology
Not really: we need an inventory of the shortcuts we have defined and implemented (i.e. using a single property instead of a longer path) so that we implement the corresponding property chain axioms. Perhaps something to discuss in Montpellier? I'm adding it to the agenda.
we need an inventory of the shortcuts we have defined and implemented
For sure, none of them are implemented. I am not sure if we defined any actually.
ACTION: Inventorier tous ces cas (à la fois par le groupe modélisation via le tableau de mapping et côté marc2rdf)
from https://docs.google.com/document/d/1cekU4np8aRWSWSKVBPQSftJOTNSPqoj8LchnDyGJXQo/edit#heading=h.d9mw6z690xu