In short, if you append Neo4jTokens.QUERY_HEADER to your value in a Neo4jIndex.get() call, it will know that you want to do a Lucene-style query. Right now, Neo4jTokens.QUERY_HEADER == "%query%" (WARNING: may change, use Neo4jTokens.QUERY_HEADER to be safe).
Its a hack.... any other ideas are more than welcome. ???
I've committed the code to GitHub and updated Blueprints 0.5-SNAPSHOT in the TinkerPop Maven2 repository with the latest deployment.
LUCA: Does OrientDB support such text searching? Should we add this too OrientGraph?