DECADES Datasets

This folder contains three groups of datasets for graph/sparse applications:

1. small - datasets that contain fewer than 1 million nodes

   • Kronecker_15.el (synthetic)
   • Amazon.tsv
   • Sinkhorn.tsv
   • YouTube.tsv

2. big - datasets that contain at least 1 million nodes

   • Kronecker_25/ (synthetic)
   • Sd1_Arc/
   • Twitter/
   • Wikipedia/

3. bipartite - datasets that are bipartite networks (partitions contain fewer than 1 million nodes)

   • Amazon.tsv
   • Dbpedia.tsv
   • Power.tsv (synthetic)
   • YouTube.tsv

The "small" and "bipartite" folders contain a single edgelist file per dataset. The "big" folder contains four binary files for each dataset (which can be parsed using Compressed Sparse Row format):

1. num_nodes_edges.txt - contains information about the size (nodes, edges) of the dataset

2. node_array.bin - binary representation of the node pointers in the dataset

3. edge_array.bin - binary representation of the edge pointers in the dataset

4. edge_values.bin - binary representation of the edge values in the dataset

We have provided a C++ script "parse_bin_files.cpp" to parse these binary files. It can be compiled and run as follows:

g++ -std=c++11 -o parse_bin_files parse_bin_files.cpp
    ./parse_bin_files [DATASET_DIRECTORY]

Permissions and Licenses

We have obtained real-world datasets from KONECT (the Koblenz Network Collection): http://konect.uni-koblenz.de/ and SNAP (Stanford Network Analysis Project): http://snap.stanford.edu/index.html. KONECT networks are licensed under the Creative Commons Attribution-ShareAlike 2.0 Germany License.