Official Implementation of Reliable Node Similarity Matrix Guided Contrastive Graph Clustering (TKDE 2024).
NS4GC has been integrated into PyAGC, a production-ready library and comprehensive benchmark for Attributed Graph Clustering. Please kindly see PyAGC Repository, PyAGC Benchmark Paper, AGC Survey Repository, and AGC Survey Paper for more details.
- PyTorch
- dgl (only as the source of datasets)
- scikit-learn
Copy hyper-parameters from params.txt to main.py and run it.
For all datasets, we use the processed version provided by Deep Graph Library.
| Dataset | Type | #Nodes | #Edges | #Features | #Clusters | Homo |
|---|---|---|---|---|---|---|
| Cora | citation | 2,708 | 10,556 | 1,433 | 7 | 82.52% |
| Citeseer | citation | 3,327 | 9,228 | 3,703 | 6 | 72.22% |
| Pubmed | citation | 19,717 | 88,651 | 500 | 3 | 79.24% |
| CoraFull | citation | 19,793 | 126,842 | 8,710 | 70 | 58.61% |
| WikiCS | reference | 11,701 | 431,726 | 300 | 10 | 65.88% |
| Photo | co-purchase | 7,650 | 238,163 | 745 | 8 | 83.65% |
| Computer | co-purchase | 13,752 | 491,722 | 767 | 10 | 78.53% |
| CoauthorCS | co-authorship | 18,333 | 163,788 | 6,805 | 15 | 83.20% |