I've built a knowledge graph based on my use case where each node (class/subclass) corresponds to a CSV file. I'm a bit confused about understanding the relationships, the source class/identifier, and the target class/identifier.
Multiple Identifiers: Should we put multiple identifiers in one box, or should we duplicate to ensure each pair is one-to-one?
Parent-Child Nodes: If one A node is the parent class of a B node, should each CSV file include at least one same column for matching or connecting them between parent and child node?
Primary Identifier: Should the primary identifier include just one factor in the box, and does this have to appear in other CSV sources besides the node CSV files?
The identifiers used are dependent on your data and your preferences. Each identifier (whether primary, source, or target) should be unique to the instance of data you are referencing.
Child inherent their parent's relationships and attributes. The parent and child definitions for classes in Designer are specific to class types. If you were building a data model for animals, you might have a parent class of Animal and a child class of Mammal. When mapping data to this data model, you might have a CSV with Dog as one of the items, Dog would traditionally be mapped to the class Mammal where it would inherit the relationships and attributes from Animal since all mammals are animals.
I would highly recommend our training courses to get started with Designer. The training will cover modeling and mapping in Designer including setting these identifiers and some best practices. We have an instructor led training starting today that goes through Wednesday -- Loading or there are also on-demand versions of these trainings available (Designer specific training -- Loading). All courses can be explored from here -- Loading.
