From the webpage:
This is a first attempt at creating a JDBC driver for the graph database Neo4j. While Neo4j is a graph database, and JDBC is based on the relational paradigm, this driver provides a way to bridge this gap.
This is done by introducing type nodes in the graph, which are directly related to the root node by the relationship TYPE. Each type node has a property “type” with its name (i.e. “tablename), and HAS_PROPERTY relationships to nodes that represent the properties that the node can have (i.e. “columns”). For each instance of this type (i.e. “row”) there is a relationship from the instance to the type node via the IS_A relationship. By using this structure the JDBC driver can mimic a relational database, and provide a means to execute queries against the Neo4j server.
Now that isn’t something you see everyday!
What if there were a GrJDBC driver? A Graph JDBC driver? Such that it views tables, rows, columns, column headers, cells, values, as graph nodes with defined properties? Read from a configuration file that identifies some database:table.
Extending the recovery of investment in large relational clusters by endowing them with graph-like capabilities (dare I say topic map like capabilities?) would be a real plus in favor of adoption. Not to mention that in read-only mode, you could demonstrate it with the client’s data.
Contrast that with all the stammering from your competition about the need to convert, etc.
I will poke around because it seems like something like that has been done but it was a long time ago. I seem to remember it wasn’t a driver but a relational database built as a graph. The same principles should apply. If I find it I will post a link (if online) or a citation to the hard copy.