#ReFeX# An implementation of Recursive Feature Extraction (ReFeX).

Described in the paper "It's Who You Know: Graph Mining using Recursive Structural Features" by Henderson et al.

Link to the paper.

Disclaimer: I have implemented this paper purely for research, educational experimentation and non-profit use. Hence doesn't come with any guarantees. Please contact the authors for any other purpose.

Running the Code:

• Requires python 2.7.

• Install package dependencies using pip install -r requirements.txt

• Run the code using python refex.py

• Run python refex.py -h for help and other program options.

• Input graph (directed graphs are supported) is specified with comma separated edge information in each line.

  • Line 1,2,3 in the input graph depicts source, destination, edge-weight.

• The epsilon equitable partition (eEP) [1] based features are also supported. The -r/--rider flag enables this.

  • The multiple eEPs are expected in a directory specified using -rd/--rider-dir.
  • Each line of the eEP file represents a cell/block of the partition, the member nodes of the cell are separated with space.
  • We compute recursive (sum and means) egonet features for eEP based primitive features.

• The test/resources directory has examples of sample input graph and epsilon equitable partitions.

• This code might (potentially) differ from the original paper, please refer to the inline text comments for more details.

• The list of primitive features and their nomenclature is captured from the original code implementation provided by the authors. We are really grateful to them for sharing the code.

[1] Pratik V Gupte and Balaraman Ravindran. "Scalable Positional Analysis for Studying Evolution of Nodes in Networks." At SIAM Data Mining Workshop on Mining Networks and Graphs: A Big Data Analytic Challenge, Philadelphia, USA, 2014.