Bipartite link prediction

Predicting links in bipartite networks on top of networkx

Node based similarities and Katz has been implemented. you can find algorithms in prediction directory. Algorithms untill now:

How to use the code

you can run the code by placing your data in inputs folder and use predict methodes iplemented in "predict" file or directly call provided functions in the evaluation file.

from bigraph import bigraph as bg

def main():
  """
  Link prediction on a bipartite network
  :return: Predicted linkes
  """
  df, df_nodes = import_files()
  G = make_graph(df)
  pr.aa_predict(G)  # Here we have called Adamic Adar method from bigraph module

or you can run evaluation methods directly which calls its peer method automatically

from bigraph.evaluation import evaluation as ev


def main():
    """
    Link prediction on a bipartite network
    :return: Predicted linkes
    """
    df, df_nodes = import_files()
    G = make_graph(df)
    ev.evaluate(G, k=10,
                method='all')  # Here we have evaluated all methods using evaluation module. Methods are 'jc', 'aa', 'pa', 'cn'

Metrics

Metrics that are calculated during evaluation:

• For further usages and calculating different metrics

Dataset format

Your dataset should be in the following format:

• Although the weight has not been involved in current version, but, the format will be the same.

More examples

from bigraph import bigraph as bg


def main():
  """
  Link prediction on a bipartite network
  :return: Predicted linkes
  """
  df, df_nodes = import_files()
  G = make_graph(df)
  pr.aa_predict(G)  # Here we have called Adamic Adar method from bigraph module
  pr.pa_predict(G)  # Prefferencial attachment
  pr.jc_predict(G)  # Jaccard coefficient
  pr.cn_predict(G)  # Common neighbors

References

TODO

• Modulate the functions
• Make it faster using vectorization etc.
• Unify and reconstruct the architecture and eliminate redundancy

Documentation

I will provide documentations whenever I could make time!:watch: or you can pull a request and help to make it happen together

1. After running the main, it will export the graph in .json and .gexf format for furthur usages. For instance: Gephi etc.

If it was helpful then hit the :star: