Published August 8, 2023 | Version v1
Dataset Open

Supplementary dataset for "Evaluating Graph Neural Networks for Link Prediction: Current Pitfalls and New Benchmarking"

Authors/Creators

  • 1. Michigan State University
  • 2. New Jersey Institute of Technology
  • 3. Snap Inc.
  • 4. Baidu Inc.

Description

The supplementary dataset for the paper "Evaluating Graph Neural Networks for Link Prediction: Current Pitfalls and New Benchmarking". We include the splits for cora, citeseer, and pubmed, the hard negative samples, and the node2vec embeddings. We also include a jupyter file read_data.ipynb to show how to read the non-txt file.

  • heart_test_samples.npy, heart_valid_samples.npy: the heard negative samples
  • *-n2v-embedding.pt: node2vec embeddings
  • test_samples_index.pt, valid_samples_index.pt: the node index of the selected samples in ogbl-ppa under HeaRT
  • gnn_feature: the input feature of cora, citeseer, pubmed

More details for our code and how to use the dataset are on the code repository: https://github.com/Juanhui28/HeaRT .

Files

read_data.ipynb

Files (2.7 GB)

Name Size Download all
md5:223f29affb86dce9e08deb586e91a4e8
2.7 GB Download
md5:88089cf5d64e0e0f5bbb7ee70a81fcfd
1.7 kB Preview Download