RDF-based inference versus non-RDF-based inference with Neo4J?

Hi Folks,

I've read that Neo4J can use RDF and add-ons for inference. But, regular queries can accomplish similar ends. It might even be possible to build up a higher-level vocabulary from basic stored procedures.

If inference is possible without adopting RDF, is there any advantage to proceeding with RDF if I'm generating my own data and schema? I'm programming in C# so cannot easily take advantage of JAVA's reasoning infrastructure (e.g. JENA).

I'd welcome any links or advice regarding inference with Neo4J, especially regarding reasoning agents.

Thank you,
-david