About the data

Node classification

Each folder such as ENZYMES consists of a set of graph files.

  • Each graph has an edge list (e.g., .edges, or .mtx).
  • Each graph also has a class label file (.nodes, or .node_labels) that encodes the class labels of the nodes. This is useful for the node classification task.



Graph classification

There is also a file in each graph collection folder (such as ENZYMES, MUTAG, etc.) that encodes the label of each graph. This is useful for the graph classification task, that is, predicting a label for each graph in the collection. The extension of the graph label file is .graphlabel, e.g., for the ENZYMES graph collection, the file is ENZYMES.graphlabel. Each line of the file encodes a label for a particular graph. For example:


1 6
2 6
3 6
4 6
5 6
...

In particular, the first line: 1 6, assigns the label 6 to the graph first graph (e.g., g1) represented by ENZYMES_g1.edges and ENZYMES_g1.nodes.

Acknowledgement policy

Please acknowledge the repository in published materials

If you publish material based on data obtained from this repository, then, in your acknowledgements, please note the assistance you received by using network repository.

Please use the following BiBTeX reference:

  @inproceedings{nr-aaai15,
      title = {The Network Data Repository with Interactive Graph Analytics and Visualization},
      author={Ryan A. Rossi and Nesreen K. Ahmed},
      booktitle = {Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence},
      url={http://networkrepository.com},
      year={2015}
  }

Many of the datasets have additional citation requests; these can be found on each dataset page. See the data license and policy for more information.