python-labelpropagation

Python implementation of label propagation

Inspired by smly's java-labelpropagation, and implemented in pure python code.

Install

python setup.py install

or

pip install labelprop

Usage

initialize labelprop

from labelprop import LabelProp

lp = LabelProp()

load data from file

lp.load_data_from_file('<FILE_NAME>')

load data from memory

lp.load_data_from_mem('<LOADED_DATA_IN_MEMORY>')

conduct label propagation

""" template

lp.run(<EPS>, <MAX_ITER>, show_log=<True/False>, clean_result=<True/False>)
    <EPS>: threshold
    <MAX_ITER>: max interation
    show_log: show report
    clean_result: if clean data

"""

# sample
ans = lp.run(0.00001, 100, show_log=True, clean_result=True) 

Data Format

Input Data Format

each line is in list format, which contains

• list[0]: node id
• list[1]: label, use 0 if unknown
• list[2]: list of neighbor nodes with weight

Example

[1, 0, [[2, 1.0], [3, 1.0]]]
[2, 1, [[1, 1.0], [3, 1.0]]]
[3, 0, [[1, 1.0], [2, 1.0], [4, 1.0]]]
[4, 0, [[3, 1.0], [5, 1.0], [8, 1.0]]]
[5, 0, [[4, 1.0], [6, 1.0], [7, 1.0]]]
[6, 2, [[5, 1.0], [7, 1.0]]]
[7, 0, [[5, 1.0], [6, 1.0]]]
[8, 0, [[4, 1.0], [9, 1.0]]]
[9, 2, [[8, 1.0]]]

Log Information

set show_log as True (default Fasle), like

lp.run(0.00001, 100, show_log=True) 

Example

Number of vertices:             0
Number of class labels:         2
Number of unlabeled vertices:   -3
Numebr of labeled vertices:     3
eps:                            1e-05
max iteration                   100

iter =  100 , eps =  8.58306884771e-06

Normal Output Data Format

set clean_result as False, like

lp.run(0.00001, 100, clean_result=False) 

each line is in list format, which contains

• list[0]: node id
• list[1]: predicted label
• list[2:]: list of labels with weight

Example

[1, 1, [1, 0.8705872816197973], [2, 0.12941033419441167]]
[2, 1, [1, 1.0], [2, 0.0]]
[3, 1, [1, 0.7411750400767577], [2, 0.2588211452259766]]
[4, 2, [1, 0.35293926912195533], [2, 0.6470559625064625]]
[5, 2, [1, 0.14117504007676362], [2, 0.8588211452259706]]
[6, 2, [1, 0.0], [2, 1.0]]
[7, 2, [1, 0.07058728161980515], [2, 0.9294103341944038]]
[8, 2, [1, 0.1764691577238195], [2, 0.8235270275789148]]
[9, 2, [1, 0.0], [2, 1.0]]

Cleaned Output Data Format

set clean_result as True, like

lp.run(0.00001, 100, clean_result=True) 

each line is in list format, which contains

• list[0]: node id
• list[1]: predicted label
• list[2:]: weight for label

Example

[1, 1, 0.999997615814209]
[2, 1, 1.0]
[3, 1, 0.9999961853027344]
[4, 2, 0.9999952316284177]
[5, 2, 0.9999961853027343]
[6, 2, 1.0]
[7, 2, 0.999997615814209]
[8, 2, 0.9999961853027344]
[9, 2, 1.0]