Bridging Nodes and Community Structures Datasets

We evaluate our framework on three real-world Ethereum money laundering datasets, including AlphahomoraExploit, CryptopiaHack, and UpbitHack. These datasets are constructed from publicly available on-chain records of major laundering incidents, covering more than 980,000 accounts and over 32,000 labeled illicit addresses in total. Each dataset contains complete transaction histories of involved addresses, with labels identifying laundering-related nodes (\u201cHeist\u201d accounts) inherited from prior studies. Compared to synthetic or small-scale benchmarks, these datasets provide realistic and large-scale transaction graphs with diverse laundering patterns, making them well-suited for evaluating detection effectiveness and network reconstruction methods.