Phishing Scam Detection on Ethereum: Towards Financial Security for Blockchain Ecosystem
Authors: Weili Chen, Xiongfeng Guo, Zhiguang Chen, Zibin Zheng, Yutong Lu
IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments show that the proposed algorithm can effectively identify phishing scams. 5 Experiment Result and Analysis |
| Researcher Affiliation | Academia | 1School of Data and Computer Science, Sun Yat-sen University 2National Supercomputer Center in Guangzhou, Sun Yat-sen University 3National Engineering Research Center of Digital Life, Sun Yat-sen University chenwli28@mail.sysu.edu.cn, guoxf6@mail2.sysu.edu.cn, zhiguang.chen@nscc-gz.cn, zhzibin@mail.sysu.edu.cn, yutong.lu@nscc-gz.cn |
| Pseudocode | Yes | Algorithm 1 The Dual-sampling Ensemble algorithm |
| Open Source Code | No | To accelerate the research in this field and promote the healthy development of blockchain technology, all relevant data and code will be released after the paper is published. |
| Open Datasets | Yes | We launch an Ethereum client, Parity4, on our server to download the ledger of Ethereum. By using Parity, we obtained all the Ethereum blocks before January 3, 2019 (to be exact, from block height 0 to block height 7,000,000). ... Fortunately, etherscan.io provides several tags for Ethereum addresses, and by crawling the website, we obtain all the addresses labeled with Phishing5. ... 4www.parity.io/ethereum/ 5etherscan.io/accounts/label/phish-hack |
| Dataset Splits | Yes | In order to reflect the effectiveness of the model more accurately and avoid the contingency caused by the partitioning of train and test sets, the paper adopts the evaluation method of k-fold cross-validation. Specifically, we set the parameter k=5. |
| Hardware Specification | No | The paper mentions running an Ethereum client 'on our server' but does not provide specific hardware details like CPU, GPU, or memory specifications. |
| Software Dependencies | No | The paper mentions 'light GBM' and the Ethereum client 'Parity' but does not provide specific version numbers for these software dependencies. |
| Experiment Setup | Yes | To compare the performance of these methods, we set the feature sampling rate to 70%, and the number of base models to 1600 (i.e., balance ensemble). Specifically, we set the parameter k=5. |