Skip to content

Karate Club: An API Oriented Open-source Python Framework for Unsupervised Learning on Graphs (CIKM 2020)

License

Notifications You must be signed in to change notification settings

tomlincr/karateclub

 
 

Repository files navigation

Version License repo size Arxiv build badge coverage badge


Karate Club is an unsupervised machine learning extension library for NetworkX.

Please look at the Documentation, relevant Paper, Promo Video, and External Resources.

Karate Club consists of state-of-the-art methods to do unsupervised learning on graph structured data. To put it simply it is a Swiss Army knife for small-scale graph mining research. First, it provides network embedding techniques at the node and graph level. Second, it includes a variety of overlapping and non-overlapping community detection methods. Implemented methods cover a wide range of network science (NetSci, Complenet), data mining (ICDM, CIKM, KDD), artificial intelligence (AAAI, IJCAI) and machine learning (NeurIPS, ICML, ICLR) conferences, workshops, and pieces from prominent journals.

The newly introduced graph classification datasets are available at SNAP, TUD Graph Kernel Datasets, and GraphLearning.io.


Citing

If you find Karate Club and the new datasets useful in your research, please consider citing the following paper:

@inproceedings{karateclub,
       title = {{Karate Club: An API Oriented Open-source Python Framework for Unsupervised Learning on Graphs}},
       author = {Benedek Rozemberczki and Oliver Kiss and Rik Sarkar},
       year = {2020},
       pages = {3125–3132},
       booktitle = {Proceedings of the 29th ACM International Conference on Information and Knowledge Management (CIKM '20)},
       organization = {ACM},
}

A simple example

Karate Club makes the use of modern community detection techniques quite easy (see here for the accompanying tutorial). For example, this is all it takes to use on a Watts-Strogatz graph Ego-splitting:

import networkx as nx
from karateclub import EgoNetSplitter

g = nx.newman_watts_strogatz_graph(1000, 20, 0.05)

splitter = EgoNetSplitter(1.0)

splitter.fit(g)

print(splitter.get_memberships())

Models included

In detail, the following community detection and embedding methods were implemented.

Overlapping Community Detection

Non-Overlapping Community Detection

Proximity Preserving Node Embedding

Structural Node Level Embedding

Attributed Node Level Embedding

Meta Node Embedding

Graph Level Embedding

Head over to our documentation to find out more about installation and data handling, a full list of implemented methods, and datasets. For a quick start, check out our examples.

If you notice anything unexpected, please open an issue and let us know. If you are missing a specific method, feel free to open a feature request. We are motivated to constantly make Karate Club even better.


Installation

Karate Club can be installed with the following pip command.

$ pip install karateclub

As we create new releases frequently, upgrading the package casually might be beneficial.

$ pip install karateclub --upgrade

Running examples

As part of the documentation we provide a number of use cases to show how the clusterings and embeddings can be utilized for downstream learning. These can accessed here with detailed line-by-line explanations.

Besides the case studies we provide synthetic examples for each model. These can be tried out by running the example scripts. In order to run one of the examples, the Graph2Vec snippet:

$ cd examples/whole_graph_embedding/
$ python graph2vec_example.py

Running tests

From the project's root-level directory:

$ pytest

License

About

Karate Club: An API Oriented Open-source Python Framework for Unsupervised Learning on Graphs (CIKM 2020)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%