README.md

jkl-serialization

Python library for (de)serializing jkl data as used by the Bayesian Network Structure Learner GOBNILP and proposed by Jaakkola et al. in Learning Bayesian Network Structure using LP Relaxations.

Installation Instructions

From PyPi

pip install jkl-serialization

(See this project on PyPi)

From Source

git clone git@github.com:daanknoope/jkl-serialization.git
cd jkl-serialization
pip install .

Example Input and Output

JKL String

3
0 4
-2.772589 2 1 2
-2.865831 0
-2.963209 1 2
-2.963209 1 1
1 4
-2.772589 2 0 2
-2.865831 0
-2.963209 1 2
-2.963209 1 0
2 4
-2.772589 2 0 1
-2.865831 0
-2.963209 1 1
-2.963209 1 0

Serialized Python Object

{
  '0': [('-2.772589', ['1', '2']),
        ('-2.865831', []),
        ('-2.963209', ['2']),
        ('-2.963209', ['1'])],
 '1': [('-2.772589', ['0', '2']),
       ('-2.865831', []),
       ('-2.963209', ['2']),
       ('-2.963209', ['0'])],
 '2': [('-2.772589', ['0', '1']),
       ('-2.865831', []),
       ('-2.963209', ['1']),
       ('-2.963209', ['0'])]
  }

Build Status