> ML_DATASET // CORA-CITATION-NETWORK_v1.0
Cora Citation Network Benchmark
McCallum et al. / Automating the Construction of Internet Portals · Graph Machine Learning & Bibliometrics · 2,708 scientific publications and 5,429 citation links across 7 machine learning sub-fields
Graph Machine Learning & BibliometricsPublic Domain / Open Academic2,708 scientific publications and 5,429 citation links across 7 machine learning sub-fieldsopen
Dataset Profile & Characteristics
Label Type:Single topic category label per paper; 1,433-dimensional binary bag-of-words features
Languages:en
License Tier:permissive-open-source
Modalities:graph, text
Intended Use
- Canonical "Hello World" benchmark for Graph Convolutional Networks (GCN, GAT)
Prohibited / Discouraged Use
- Extrapolating performance to massive industrial heterophilous graphs
Bias, Leakage & Privacy Risk Analysis
Privacy / Sensitive Data Risks:
Academic research papers and citation graphs; zero personal data.
Known Bias:
Small graph size (2,708 nodes); high homophily (connected nodes share same label).
Known Benchmark Leakage:
Standard Planetoid semi-supervised split (140 training nodes).