This repository provides a reference implementation of struc2vec as described in the paper:

struc2vec: Learning Node Representations from Structural Identity.
Leonardo F. R. Ribeiro, Pedro H. P. Saverese, Daniel R. Figueiredo.
Knowledge Discovery and Data Mining, SigKDD, 2017.

The struc2vec algorithm learns continuous representations for nodes in any graph. struc2vec captures structural equivalence between nodes.

Before to execute struc2vec, it is necessary to install the following packages:
pip install futures
pip install fastdtw
pip install gensim

To run struc2vec on Mirrored Zachary's karate club network, execute the following command from the project home directory:
python src/main.py --input graph/karate-mirrored.edgelist --output emb/karate-mirrored.emb

To activate optimization 1, use the following option: --OPT1 true
To activate optimization 2: --OPT2 true
To activate optimization 3: --OPT3 true

To run struc2vec on Barbell network, using all optimizations, execute the following command from the project home directory:
python src/main.py --input graph/barbell.edgelist --output emb/barbell.emb --num-walks 20 --walk-length 80 --window-size 5 --dimensions 2 --OPT1 True --OPT2 True --OPT3 True --until-layer 6

You can check out the other options available to use with struc2vec using:
python src/main.py --help

The supported input format is an edgelist:

node1_id_int node2_id_int

The output file has n+1 lines for a graph with n vertices. The first line has the following format:

num_of_nodes dim_of_representation

The next n lines are as follows:

node_id dim1 dim2 ... dimd

where dim1, ... , dimd is the d-dimensional representation learned by struc2vec.

Please send any questions you might have about the code and/or the algorithm to [email protected].

Note: This is only a reference implementation of the framework struc2vec.