Demo

sh demo.sh

• Results in toy.inferred and toy.result

Usage

python socnl.py [training filepath] [graph filepath] [# of nodes] [# of labeled nodes] [# of labels] [lambda] > [inferred filepath]

python evaluation.py [test filepath] [inferred filepath] > [result filepath]

• socnl.py requires numpy/scipy

File format

Make sure to start all node ids from 0.

Graph file format

[src node id] \t [dst node id]

• See toy data file

Training file format

[node id] \t [label id]

• Make sure to list labeled nodes on top, and unlabeled nodes on bottom.
• For unlabeled nodes, let [label id] = -1.
• See toy data file

Test file format

[node id] \t [label id]

• Make sure to list only test nodes.
• See toy data file

Inferred file format

[inferred label id] \t [confidence value]

Result file format

[precision@p] [p]