Intermediate Node Modeling Question


(Benjamin Pope) #1

Apologies in advance if this topic has been beat to death but after going through the beginner tutorials, I have a modeling question and in particular how, the fact that Nodes have to be unique, impacts the model.

Background: two different DICTIONARIES, two DEFINITIONS for the same TERM "Aardvark"
My brute force model doesn't work since you can't tell which DEFINITION belongs to which DICTIONARY
(a) The first 'workable' solution seems to be valid but something doesn't seem very elegant about it.
(b) The second 'workable' solution seems to better - if the property is on the node (since it's an adjective) but does mean that property will be duplicated in many nodes?
(c) Then I found a video on Intermediate Nodes and since it seems like my original problem would be modeled as illustrated in the Not Valid image but could be implemented as shown in the last graphic.

I was wondering if my analysis "made sense" to folks with actual experience ;-) and if I've applied the concept of intermediate nodes correctly?


-Thanks, Ben


(Jiropole) #2

Hi Ben,

I think you can avoid use of intermediate nodes. You know that a node may have any number of relationships, incoming or outgoing? In any case, I would say your solution (2) is the best option.

If you got crazy and added more languages, now you're more in the territory of intermediate nodes. In that case, your Definition nodes would be almost empty, but would connect to several "LocalizedDefinition" nodes that contain language specific def values.