ehb2126 / psl

The PSL software from the University of Maryland and the University of California Santa Cruz

PSL

Build Status

Probabilistic soft logic (PSL) is a machine learning framework for developing probabilistic models. PSL models are easy to use and fast. You can define models using a straightforward logical syntax and solve them with fast convex optimization. PSL has produced state-of-the-art results in many areas spanning natural language processing, social-network analysis, knowledge graphs, recommender system, and computational biology. More information about PSL is available at the PSL homepage.

Getting Started with PSL

If you want to use PSL to build models, you probably do not need this source code. Instead, visit the Getting Started guide to learn how to create PSL projects that will automatically install a stable version of these libraries.

Installing PSL from Source

If you do want to install PSL from source, you can use Maven 3.x. In the top-level directory of the PSL source (which should be the same directory that holds this README), run:

mvn install

Installing PSL with Gurobi

PSL can additionally be used with the Gurobi solver for inference. Gurobi is a commercial solver, but free academic licenses are available. To use Gurobi with PSL, you must have Gurobi installed and licensed, see Gurobi Quickstart Guide. Further, you must install the Gurobi jar file into your local Maven repository. See Guide to installing 3rd party JARs for more information.

To do this, first download the Gurobi jar file from the Gurobi website. You will need to create an account and agree to the license terms. You must also obtain a Gurobi license that is registered and saved to your machine. Be sure to export the GUROBI_HOME environment variable to point to your install directory, <installdir>, and GRB_LICENSE_FILE environment variable to point to the location of the license file. Moreover, you must have the Gurobi install bin directory, <installdir>/bin, added to your PATH environment variable and <installdir>/lib added to your LD_LIBRARY_PATH environment variable. Then, run the following command, replacing <installdir>/lib/gurobi.jar with the path to the downloaded jar file and <version> with the version of Gurobi you downloaded:

mvn install:install-file -Dfile=<installdir>/lib/gurobi.jar -DgroupId=com.gurobi -DartifactId=gurobi -Dversion=<version> -Dpackaging=jar

If you are using a version of Gurobi other than 10.0.3, you will also need to update the Gurobi dependency version in the PSL pom.xml file. Then, you can install PSL with Gurobi support by running:

mvn install -P Gurobi

PSL inference can then be run with Gurobi using the GurobiInference class.

Citing PSL

We hope you find PSL useful! If you have, please consider citing PSL in any related publications as

@article{bach:jmlr17,
  Author = {Bach, Stephen H. and Broecheler, Matthias and Huang, Bert and Getoor, Lise},
  Journal = {Journal of Machine Learning Research (JMLR)},
  Title = {Hinge-Loss {M}arkov Random Fields and Probabilistic Soft Logic},
  Year = {2017}
}

Additional Resources

• PSL Homepage
• PSL Examples
• API Reference
• PSL Source Repository
• PSL Wiki
• Getting Started Guide
• User Group