nlpub/chinese-whispers

An implementation of Chinese Whispers in Python.

Chinese Whispers for Python

This is an implementation of the Chinese Whispers clustering algorithm in Python. Since this library is based on NetworkX, it is simple to use.

Installation

• pip: pip install chinese-whispers
• Anaconda: conda install -c conda-forge chinese-whispers
• Mamba: mamba install -c conda-forge chinese-whispers

Usage

Given a NetworkX graph G, this library can cluster it using the following code:

from chinese_whispers import chinese_whispers
chinese_whispers(G, weighting='top', iterations=20)

As the result, each node of the input graph is provided with the label attribute that stores the cluster label.

The library also offers a convenient command-line interface (CLI) for clustering graphs represented in the ABC tab-separated format (source\ttarget\tweight).

# Write karate_club.tsv (just as example)
python3 -c 'import networkx as nx; nx.write_weighted_edgelist(nx.karate_club_graph(), "karate_club.tsv", delimiter="\t")'

# Using as CLI
chinese-whispers karate_club.tsv

# Using as module (same CLI as above)
python3 -mchinese_whispers karate_club.tsv

A more complete usage example is available in the example notebook and at https://nlpub.github.io/chinese-whispers/.

In case you require higher performance, please consider our Java implementation that also includes other graph clustering algorithms: https://github.com/nlpub/watset-java.

Citation

• Ustalov, D., Panchenko, A., Biemann, C., Ponzetto, S.P.: Watset: Local-Global Graph Clustering with Applications in Sense and Frame Induction. Computational Linguistics 45(3), 423–479 (2019)

@article{Ustalov:19:cl,
  author    = {Ustalov, Dmitry and Panchenko, Alexander and Biemann, Chris and Ponzetto, Simone Paolo},
  title     = {{Watset: Local-Global Graph Clustering with Applications in Sense and Frame Induction}},
  journal   = {Computational Linguistics},
  year      = {2019},
  volume    = {45},
  number    = {3},
  pages     = {423--479},
  doi       = {10.1162/COLI_a_00354},
  publisher = {MIT Press},
  issn      = {0891-2017},
  language  = {english},
}

Copyright

Copyright (c) 2018–2024 Dmitry Ustalov. See LICENSE for details.

chinese-whispers's Issues

Question about the algorithm

Hello,

I have a question about the Chinese Whisper (CW) algorithm. How does it differ from the Label Propagation Algorithm (LPA)? which is implemented by NetworkX.

When should I use CW? and when LPA?

I'm working on a paper about community detection. and I appreciate the help.

Recieving error 'Graph' object has no attribute 'node'

I am working on a nlp project, and I used the chinese-whispers package. But I encountered this error:

File "/home/usr/.local/lib/python3.6/site-packages/chinese_whispers/init.py", line 44, in chinese_whispers
G.node[node]['label'] = i + 1
AttributeError: 'Graph' object has no attribute 'node'

where is python file with code?

great idea , but there is py file with code?

Conda forge package

Would it be possible to install chinese-whispers through anaconda? Perhaps through conda-forge?

Is there any way to specify the number of labels and the number of members under each label

Hi!
I have been giving the python code on https://github.com/nlpub/chinese-whispers-python a test. I was just wondering if there is any way to specify the total number of labels, and the number of members under each label, or roughly so. Say, I have 1000 members, and I want to cluster them into 200 groups, allowing 1 to 10 member in each group, 1 meaning that member is a group by himself.

Thanks,
Wei