You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

README.txt 2.5KB

4 years ago
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475
  1. README for dataset DD
  2. === Usage ===
  3. This folder contains the following comma separated text files
  4. (replace DS by the name of the dataset):
  5. n = total number of nodes
  6. m = total number of edges
  7. N = number of graphs
  8. (1) DS_A.txt (m lines)
  9. sparse (block diagonal) adjacency matrix for all graphs,
  10. each line corresponds to (row, col) resp. (node_id, node_id)
  11. (2) DS_graph_indicator.txt (n lines)
  12. column vector of graph identifiers for all nodes of all graphs,
  13. the value in the i-th line is the graph_id of the node with node_id i
  14. (3) DS_graph_labels.txt (N lines)
  15. class labels for all graphs in the dataset,
  16. the value in the i-th line is the class label of the graph with graph_id i
  17. (4) DS_node_labels.txt (n lines)
  18. column vector of node labels,
  19. the value in the i-th line corresponds to the node with node_id i
  20. There are OPTIONAL files if the respective information is available:
  21. (5) DS_edge_labels.txt (m lines; same size as DS_A_sparse.txt)
  22. labels for the edges in DS_A_sparse.txt
  23. (6) DS_edge_attributes.txt (m lines; same size as DS_A.txt)
  24. attributes for the edges in DS_A.txt
  25. (7) DS_node_attributes.txt (n lines)
  26. matrix of node attributes,
  27. the comma seperated values in the i-th line is the attribute vector of the node with node_id i
  28. (8) DS_graph_attributes.txt (N lines)
  29. regression values for all graphs in the dataset,
  30. the value in the i-th line is the attribute of the graph with graph_id i
  31. === Description ===
  32. D&D is a dataset of 1178 protein structures (Dobson and Doig, 2003). Each protein is
  33. represented by a graph, in which the nodes are amino acids and two nodes are connected
  34. by an edge if they are less than 6 Angstroms apart. The prediction task is to classify
  35. the protein structures into enzymes and non-enzymes.
  36. === Previous Use of the Dataset ===
  37. Neumann, M., Garnett R., Bauckhage Ch., Kersting K.: Propagation Kernels: Efficient Graph
  38. Kernels from Propagated Information. Under review at MLJ.
  39. Neumann, M., Patricia, N., Garnett, R., Kersting, K.: Efficient Graph Kernels by
  40. Randomization. In: P.A. Flach, T.D. Bie, N. Cristianini (eds.) ECML/PKDD, Notes in
  41. Computer Science, vol. 7523, pp. 378-393. Springer (2012).
  42. Shervashidze, N., Schweitzer, P., van Leeuwen, E., Mehlhorn, K., Borgwardt, K.:
  43. Weisfeiler-Lehman Graph Kernels. Journal of Machine Learning Research 12, 2539-2561 (2011)
  44. === References ===
  45. P. D. Dobson and A. J. Doig. Distinguishing enzyme structures from non-enzymes without
  46. alignments. J. Mol. Biol., 330(4):771–783, Jul 2003.