Python Relational Data Mining

This python project was created to enable easier use of several inductive logic programming (ILP) and relational data mining (RDM) algorithm implementations. One important aim of the project is to offer a common bridge between a RDBMS and the ILP&RDM implementations, since many of approaches accept databases in their own format.

This project also includes the UI components (widgets) for the ClowdFlows data mining platform.

Currently, the project offers support for MySQL, PostgreSQL, SQLite and CSV files-based databases and the following algorithms: Aleph, RSD, Wordification, TreeLiker, Caraf, Relaggs, Quantiles, Cardinalization, 1BC, 1BC2, and Tertius.

Included approaches

Although python-rdm itself is MIT licensed, we include approaches that have their own licenses (all of the sources are unmodified). To be sure, please contact the respective authors if you want to use their approach for any commercial purposes.

• Aleph
  • Official page
  • Freely available for academic purposes, contact the author Ashwin Srinivasan for commercial use
  • The source code is included here (aleph.pl)
• RSD
  • by Filip Železný et al
  • Official page
  • The source code is included here (.pl files)
  • Included with permission by the author
• TreeLiker (includes HiFi, RelF and Poly)
  • Official page
  • The binaries are included here
  • GPL license
• Wordification
  • by Matic Perovšek et al
  • python-rdm is currently the main repository for this approach.
  • The source code is included here
  • MIT license

Nicolas Lachiche's team at the University of Strasbourg contributions:

• 1BC, 1BC2, Tertius
  • By Peter Flach and Nicolas Lachiche
  • Sources included here here
  • Official sites: Tertius, 1BC
  • Included with permission by the authors; please contact the authors for commercial use
• Caraf
  • By Clement Charnay, Agnès Braud and Nicolas Lachiche et al
  • All implemented in the Caraf java binaries included here
  • Included with permission by the authors; please contact the authors for commercial use
• Relaggs (Krogel and Wrobel, 2001), Quantiles, Cardinalization
  • Original Proper adapted by Nicolas Lachiche et al
  • GPLv2 license
  • All implemented in the Proper java binaries included here

Installation, documentation

Please find installation instructions, examples and API reference on Read the Docs.

Note

Please note that this is a research project and that drastic changes can be (and are) made pretty regularly. Changes are documented in the CHANGELOG.

Pull requests and issues are welcome.

Contributors to the RDM package code

Anže Vavpetič (@anzev), Nicolas Lachiche, Alain Shakour (@alshak), Matic Perovšek (@mperice), Vid Podpečan (@vpodpecan)

• Knowldge Technologies Department, Jožef Stefan Institute, Ljubljana
• Engineering, Computer and Imaging Sciences Laboratory, University of Strasbourg, France