Skip to content

Latest commit

 

History

History
20 lines (16 loc) · 1.71 KB

README.md

File metadata and controls

20 lines (16 loc) · 1.71 KB

Phishing Detection on Ethereum

Introduction

Recently, blockchain technology has become a topic in the spotlight but also a hotbed of various cybercrimes. Ethereum is currently the largest blockchain platform that supports smart contracts and the corresponding cryptocurrency ether is the second-largest cyptocurrency. Besides, among various security issues of blockchain digital cryptocurrency, the number of phishing scams accounts for more than 50% of all cybercrimes in Ethereum since 2017 and this kind of scam has become as a main threat to trading security of Ethereum, thus emerging as a serious threat to the trading security of the blockchain ecosystem.

Our work shares phishing account information from Etherscan and the code for how to crawl it. In addition, the trans2vec algorithm for detection was also shared.

Dataset

Labeled phishing account

We crawled the phishing accounts before January 3, 2019 and their first-order nodes from https://etherscan.io/accounts/label/phish-hack, and got a total of 1259 phishing accounts. We call these phishing accounts as positive examples.

Ethereum transaction network

As 1259 addresses are labeled as phishing nodes which are the targets of the detection approach, we randomly select 1259 unlabeled nodes as the outliers. With these labeled and unlabeled nodes being the central nodes, we extract their first-order neighbors and the con- nected edges between all of them to form a subnetwork.

Code

The crawler code

We provide the code to crawl information from https://etherscan.io/.

Trans2vec

We propose a novel network embedding model trans2vec specifically for transaction networks by incorporating the transaction amount values and timestamps of transaction links.