This demo Java command-level application loads an embedded Neo4j graph database with publicly available railroad locations. A graph database consists of nodes connected by links. Nodes and links can have name / value properties. The DBMS provides an API to navigate the resulting graph. A Lucene index enables an application to find a node-of-interest.
In our case, each node is a integer-numbered "station" from a subset of public 1996 census data of the USA railroad network. A link (track) connects two stations. Each link has a node-to-node distance property. Neo4j has a built-in A* shortest path algorithm that returns an iterator across the shortest path through the graph of "stations". Our Command executable class accepts a "from" and "to" integer pair (1..133752) from the command line, to emit Google KML for the shortest path between the two stations. It pipes the KML to a local Google Earth executable if found.
Clone this repository to your file system. We used Eclipse 3.7 Juno. The directory tree is suitable to import into Eclipse as a Maven project. You may need environmental adjustments if your computer is not a Mac on Mountain Lion.
A Java 1.6 compiler level assumed throughout. You can run these classes directly from Eclipse. Use a Run Configuration to supply the "from" and "to" integers.
- Reloads the Neo4j graphdb from CSV data located on the class path
- Only required if you are curious, or if new you obtain newer CSV input.
- Displays the Lucene index of nodes (stations) from the graphdb.
- The index is used to find the "from" and "to" nodes to supply to the built-in A* algorithm.
- This class is for the curious -- it is long-running, pegging CPU cores, running the fan.
- This is the demonstration goal.
- The From and to station IDs are integers taken from the range 1..133752.
- The application displays a help blurb if it finds no input pair.
- The application ignores bad input.
- A route displays as a white line -- Google Earth zooms to accommodate.