Getting started

Installation

Step 0: Make sure you have all the required components installed; in particular, the Java runtime environment, Apache Maven and Conda (or a Python 2.7 installation).

Step 1: Configure the Python virtual environment via conda

  conda env create -f environment.yml

and subsequently activate the environment. (Alternatively, have a Python 2 installation ready, the python executable being your PATH, with the required dependencies listed in environment.yml.)

Step 1: Go to the directory where cloned ProbCog or unpacked the archive you downloaded to an empty directory of your choice. We shall refer to the directory as /path/to/probcog.

  cd /path/to/probcog

Step 2: Configure the Maven POM for your platform

  python configurePOM.py

Step 3: Compile the Java sources using Maven.

  mvn compile

Step 4: Create the ProbCog applications by running the make_apps.py script:

  python make_apps.py

This will create a number of executable shell scripts (or executable .bat files on Windows) in the /path/to/probcog/apps directory.

Step 5: Configure your environment

When make_apps has finished creating the applications, it will tell you how to configure your environment. Simply follow the instructions that are given.

First Steps

• To get a first impression, open a shell and go to /path/to/probcog/examples/alarm. Then run blnquery and mlnquery from there; these are the graphical inference tools for Bayesian logic networks (BLNs) and Markov logic networks (MLNs) respectively.
• Familiarize yourself with the ProbCog tools and check out the various tutorials