Machine learning tools for massively multi-labeled networks

You can install this library by running the following command inside the top-level directory

pip install .

The mmln library represents networks using networkx. Label information is stored in dictionaries using three node attributes, mmln.OBSVS, mmln.TARGETS, and mmln.TRUTH. mmln.OBSVS stores label values that are observed, i.e. given. For example,

network.node['Node 1'][mmln.OBSVS] = {'Label 1': 1, 'Label 2': 0}

says that Label 1 on Node 1 is observed with a value of 1, i.e. True and Label 2 is observed with a value of 0, i.e. False.

Likewise, mmln.TARGETS stores predictions, such as from a prediction algorithm. Putting an entry for a label in a mmln.TARGETS dictionary attribute indicates to prediction algorithms that they should make a prediction for that node-label pair. The initial value stored for the target does not matter. Prediction algorithms will overwrite it.

mmln.TRUTH is for storing true values for the corresponding targets on the node.