Control mapping to nodes vs properties in Neosemantics

[nicolaiwadstrom]

It would be valuable to be able to map RDF triples with literal values (vs Resource references) to graph nodes (vs properties), such that:

Triple:
Subject: ns:SomeId
Predicate: rdfs:label
Object: "Hello World, I am a string property!"

Would map to

Vertex[ns:SomeId] ------Edge[rdfs:label]------> Vertex[ns:xxxx]

and Vertex[ns:xxxx] has a property called value that equals "Hello World, I am a string property!"

vs the current option where RDF literals are always mapped to properties on Vertex[ ns:SomeId ]

The rational for this is that it would allow multiple edges being expressed from various data sources (which is common in RDF data), and also additional properties to be attached to the relationships (source, authority etc).

Preferable this should be another configuration option in NeoSemantics (just handleRDFTypes configuration property).

In our company, our primary knowledge graph models use this type of mapping as outlined above. We are porting some of our knowledge graph systems to support Neo4j and exploring various options including being compatible with NeoSemantics.

For us this model have been bridging the divide between Triple based store models and property graphs in an effective manner for building knowledge graph systems.

There are other types of knowledge graph systems that use similar models, or even more complex relationships (Vaticle's TypeDB for example - even though they do not follow an RDF model).

Question is would others value support for this?