twicerogue / sbm_node_embedding Goto Github PK

This repository contains all the functions and scripts for getting word2vec embeddings for community detection on graphs. For details please refer to the following paper:

W Ding, C Lin, P Ishwar, "Node Embedding via Word Embedding for Network Community Discovery", in arXiv preprint arXiv:1611.03028, 2016.

A shorter version of this is in IEEE ICASSP conference (2017).

The functions in this repository requires only basic python packages (scipy, numpy, sklearn, and gensim). No additional installation is required.

The functions to create embeddings and communities are in VEClib.py and utils.py.

Please check the example code in /example folder to see how these functions can be used for community detection.

We provide two example scripts in /example folder to demo how to use our code for community detection for simulated graphs, as well as real word graphs.

NOTE: /all_script folder has all the scripts for re-creating the plots in the abovementioned paper where we demonstrates the empirical performance around the information theory limits of Stochastic Block Models.