Skip to main content

Add a short description here!

Project description


Tool to convert datasets from Benchmark Data Sets for Graph Kernels (K. Kersting et al., 2016) into a format suitable for deep learning research in graph classification.


Simply run pip install dortmund2array to install the command-line interface. The only dependencies are numpy networkx pandas.


Given any benchmark dataset, this tool will create a file DATASET.pickle that contains a pickled list. At index i the list has a dictionary with the adjacency matrix, the graph signal (also known as graph feature matrix) and the corresponding label for the ith graph.

    "adjacency":    ...  # as numpy array. Shape: (nodes, nodes)
    "graph_signal": ...  # as numpy array. Shape: (nodes, features)
    "label":        ...  # usually a scalar.

The graph signal is an array of shape (nodes, features) where the features are either attributes given by the dataset or if no attributes are available, we simply take the node labels as attributes.

How to use

Simply get the dortmund2array command line tool via pip install dortmund2array.

usage: dortmund2array [-h] [--version] [--output OUTPUT_FOLDER]
                      [--input INPUT_FOLDER]

Tool to convert datasets from 'Benchmark Data Sets for Graph Kernels' (K.
Kersting et al., 2016)

optional arguments:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
                        Output folder.
                        Input folder containing the dataset of the same name.

Thus, download and unzip a dataset. Make sure the folder-name agrees with the dataset-name on the files inside of it. If you for instance download MUTAG and the corresponding folder is .\MUTAG\ and you want the array data saved in .\MUTAG_array\ then you need to simply run:

dortmund2array -i ./MUTAG/ -o ./MUTAG_array/


Make sure you meet all the dependencies inside the requirements.txt.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dortmund2array-0.3.1.tar.gz (15.4 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page