FlowScope: Spotting Money Laundering Based on Graphs
Authors: Xiangfeng Li, Shenghua Liu, Zifeng Li, Xiaotian Han, Chuan Shi, Bryan Hooi, He Huang, Xueqi Cheng4731-4738
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Flow Scope outperforms state-of-the-art baselines in accurately detecting the accounts involved in money laundering, in both injected and real-world data settings. and Experiments The datasets used are summarized in Table 2, CBank dataset: Real-world transfer data from an anonymous bank under an NDA agreement, with a group of money laundering accounts being labeled, and the accounts opened in other banks are labeled. |
| Researcher Affiliation | Collaboration | 1Beijing University of Post and Telecommunication 2Institute of Computing Technology, Chinese Academy of Sciences 3University of Surrey, 4Texas A&M University 5School of Computer Science, National University of Singapore 6China Citic Bank |
| Pseudocode | Yes | Algorithm 1: Flow Scope |
| Open Source Code | Yes | Our algorithm is reproducible, and open-sourced 1. 1See code in https://github.com/aplaceof/Flow Scope |
| Open Datasets | Yes | Czech Financial Dataset (CFD): An anonymous transfers of Czech bank released for Discovery Challenge in PKDD 99 (L utkebohle ). |
| Dataset Splits | No | The paper describes how synthetic data is generated and injected for experiments and how real-world data is used, but it does not specify explicit train/validation/test dataset splits with percentages or counts for reproducibility. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU models, CPU types, or memory specifications used for running the experiments. |
| Software Dependencies | No | The paper mentions the use of "SNAP network library" but does not provide specific version numbers for it or any other software dependencies, libraries, or frameworks used in the experiments. |
| Experiment Setup | Yes | We inject ML as follows: fraudulent accounts are randomly picked from V as the multipartite groups, e.g. A, M and C. Denote the total number of injected accounts as N. The edges between each group are randomly generated with probability p. The total amount of laundering money D, starting from A, through M, to C, is assigned to generate edges proportional to Gaussian distribution (mean = 10 and std = 1). and We hence set λ to 4.0 for all our experiments as we assume 1/λ= 0.25 is the highest rate of retention or deficit as camouflage that fraudsters can afford. |