CryptoGraphNet

Clean repository package for the CryptoGraphNet link-prediction project, including:

• the custom Bitcoin transaction dataset
• the public Elliptic Bitcoin dataset
• the benchmark runners for both datasets
• the standalone NOS heuristic baseline script
• figure-generation and environment setup scripts
• the updated LaTeX manuscript project
• the reviewer-response source files

Repository Layout

CryptoGraphNet_GitHub_Clean/
├── datasets/
│   ├── custom/
│   │   └── Dataset.parquet
│   └── elliptic/
│       ├── elliptic_txs_classes.csv
│       ├── elliptic_txs_edgelist.csv
│       └── elliptic_txs_features.csv
├── docs/
│   ├── graph_lp_comparison.docx
│   ├── paper_math_and_algorithm.md.pdf
│   ├── latex/
│   │   ├── main.tex
│   │   ├── README.md
│   │   └── figures/
│   └── review/
│       ├── response_to_reviewers.tex
│       ├── Response_to_Reviewers_CryptoGraphNet.docx
│       └── Response_to_Reviewers_CryptoGraphNet.md
├── scripts/
│   ├── setup_env.py
│   ├── run_benchmark.py
│   ├── run_benchmark_elliptic.py
│   ├── compute_nos_heuristic_baseline.py
│   └── paper_figures.py
├── requirements.txt
├── .gitignore
└── .gitattributes

Datasets

• datasets/custom/Dataset.parquet: custom Bitcoin transaction dataset used by the main benchmark.
• datasets/elliptic/: public Elliptic dataset used by the separate Elliptic runner.

Because these dataset files are much larger than normal GitHub size limits, this repository is prepared for Git LFS.

Quick Start

Install dependencies:

python scripts/setup_env.py

Run the custom-dataset benchmark:

python scripts/run_benchmark.py --stage all

Run the Elliptic benchmark:

python scripts/run_benchmark_elliptic.py --stage paper --seeds 42

Run the standalone NOS heuristic baseline:

python scripts/compute_nos_heuristic_baseline.py

Generate paper figures:

python scripts/paper_figures.py

GitHub / LFS Notes

Before pushing to GitHub, enable LFS in this repository:

git lfs install

The .gitattributes file already marks the dataset files for LFS tracking.

Generated Artifacts

Benchmark outputs and caches are intentionally excluded from version control. When you run the scripts, they will create local folders such as:

• benchmark_results/
• benchmark_checkpoints/
• benchmark_results_elliptic/
• benchmark_checkpoints_elliptic/
• figures/

LaTeX build intermediates are also excluded. The updated manuscript project is available under docs/latex/, and the reviewer-response materials are available under docs/review/.