The software presented hereafter have been developed during the PhD Thesis of Quang-Thang Dinh. They consist of a set of Markov Logic Network learning tools, that rely on the Alchemy library.


  • Generative Structure Learning for MLN based on Graph of Predicates.

  • Source codes (C++) comming soon for [Download]


  • Discriminative MLN Structure learning based on Propositionalization.

  • Source code (C++) comming soon for [Download]


  • Heuristic Generative Structure learning for MLNs

  • Source codes (C++) comming soon for [Download]


  • Heuristic Generative Structure learning for MLNs.

  • Source codes (C++) comming soon for [Download]

Instructions for Running

These systems are implemented over Alchemy. These instructions support to incorporating HGSM, HDSM and DMSP into the alpha version of Alchemy (Oct. 22, 2007) and GSLP into the beta version of Alchemy.

Download and install Alchemy, the alpha version (Oct. 22, 2007) for HGSM, HDSM and DMSP and the beta version for GSLP.
Download our source codes (HGSM, HDSM, DMSP or GSLP), unpack it and copy all source files (excluding the makefile and the readme.txt) to the src/learnstruct folder of the Alchemy source.
Copy the makefile to the src folder of Alchemy.
Type make depend
make hgsm to compile HGSM,
make hdsm to compile HDSM,
make dmsp to compile DMSP,
make gslplc; make gslplw; make gslpprune to compile GSLP.
Read the readme.txt file to see how to use each system to learn a MLN structure.