The JAM Project



How to Run JAM 2.11


How to Run

  1. Verify installation is complete
    Before running the JAM program, make sure that JDK1.1 is installed, and CLASSPATH point at the JDK-2.11/src directory.

  2. Verify that the configuration is complete for the server and all clients
    Before running the JAM program, actually for the server and clients to start successfully, you must create configuration files for the server and all clients. One of the function's of the server's configuration file is security -- the valid client (data sites) which the server (Configuration Manager) is assigned the task of managing. Configuration Directions

    Note: Create the meta-learning attribute file, using the name: ``tmpcom.attr''. It is the dictionary file that describes the meta-learning training set, (i.e. the correct label, and predictions of the local and remote classifiers). See the demo/crx/mango directory for an example

  3. Start the Configuration Manager

    On the designated server machine:

    1. At the prompt type:
      $ java jam.cfmgr.ConfigFileMgr [portNumber] [number of data sites]

    After 10 seconds of start-up time, the Configuration Manager appears.

  4. Start each client (Data Site)

    On each designated client machine:

    1. Go to the client's home directory.
    2. At the prompt type:
      $ java jam.main.JAM [options]
      where options is: -[tigs] and
      • t: for the text base JAM
      • i: interactive mode - get user input
      • g: use DotServer for graph construction (if dot.exe is not available)
      • s: for a run in a single mode (i.e. no meta-learning)

    After 10-15 seconds of start-up time, the Data Site screen appears. The initialized state is unconnected, meaning that, by default, the data site is not connected to the server. You will connect to the network once you press the start button.

    Why is this step necessary? Because this allows the user to manipulate the local data prior to establishing the connection. This design decision also enables the Data Site program to be used solely as a front-end to the various available machine learning algorithms.


    Columbia University, Sept 1997. Last Modified: June 5, 1998
    andreas@cs.columbia.edu