Home

ProPPR: Programming with Personalized PageRank

Publications

• [Programming with Personalized PageRank: A Locally Groundable First-Order Probabilistic Logic] (http://www.cs.cmu.edu/~wcohen/postscript/cikm-2013-proppr.pdf) CIKM-2013, Honorable Mention for Best Paper

Beyond the README

• Implementation Details: Parameter Vector and other tricky components
• Walkthroughs and Tutorials: Line-by-line demos
• Concepts: What's a logic program? What's a graph? What's being learned? What are the features?
• Using Features in ProPPR: How to pair the right feature with the right program
• Best Practices: How do I build a good dataset?
• Troubleshooting: Common error messages and how to diagnose them

Contributing to ProPPR

• Commit Checklist: How to run tests and avoid breaking the build
• API Use Cases: In theory, when we make something possible in the API, we'll document it here.

Version History

• Initial python prototype, Winter 2012
• Initial Java learner, Winter 2012/2013
• Initial Java prover, Spring 2013
• Proving revision of python prototype, Summer 2013 (newcomponents etc)
• Full Java system, Autumn 2013?
• WAM python prototype, Summer 2014
• WAM Java system, Autumn 2014 (branch:2.0)
• Optimized SRW math, Winter 2014/2015 (branch:srw2.0)