This is a general, “best practice” kind of question and in the context of modeling with CIDOC-CRM.
I understand that if I model a relationship with a property in a template, that relationship will show up for the item using the template in the property field, and for the linked item, in the Linked Resources. I am trying to come up with some basic pattern for when I use which type of relationship (and why).
For example, in our data we often have hierarchies … for example of administrative units, where a region has a 1:M “parent:child” relationship to several subregions. On the Administrative Unit template I could have two properties (e.g. falls within and contains) that will then show as properties both the parent and the children of a viewed item. Or – I could just have the “falls within” or the “contains” and let the missing relationship property then appear as Linked Resources.
This has implications for how site visitors would perceive the data as the relationships indicated by Linked Resources could be positioned on the page separately (or left off entirely) from the relationships that are contained in properties in the values block.
Is the best practice for the 1:M relationship to emulate a normalized data structure (child has link to parent) or some other suggested way to think about this?
Similarly for M:M types of relationships, is the best practice to use Inverse Properties (or otherwise manually create relationships) so that there is a property in each resource template?
Somewhat related is that I cannot use the same property twice in a resource template. So for example: P89 Falls Within which relates two instances of E53 Place, with many E53 items falling within a single E53 item. If my current Place Item happens to fall within two conceptually distinct other Place Items (e.g. an Administrative Unit and a local Toponym modeled as Locality), I cannot use P89 twice with different Alternate Names (here Administrative Unit and then Locality). In these cases, it seems I have to find another relationship that sort of works, like P53 has former or current location. Am I missing anything here?
Thanks for any insight!