Implementation of Deep Neural Network in the Design of Ethereum Blockchain Scam Token Detection Applications

Abstract

The popularity of blockchain continues to increase as technology develops, especially in the context of Ethereum as one of the leading blockchain platforms. However, this increase was also followed by many cases of fraud, especially in the form of tokens. In blockchain technology, tokens often refer to cryptocurrencies or digital currencies used as a means of exchange related to a particular project or platform. This research designs and builds an application system that can detect scam crypto tokens on the Ethereum blockchain, specifically for the ERC-20 (Ethereum Request for Comments 20) token type, which was proposed by Fabian Vogelsteller in November 2015, is a token standard that implements APIs for tokens. in Smart Contracts. Making a scam detection application implements the deep learning method with the Deep Neural Network (DNN) algorithm and evaluates performance using two test scenarios by dividing the dataset into three ratios of training data and test data. The output of the application is JSON-RPC which is integrated with the website. In testing the DNN model, using 80% training data and 20% test data, the DNN algorithm provides an accuracy of 0.997558%. Furthermore, system testing was carried out involving various scenarios to verify its functionality, including input validation, data extraction, DNN prediction, and display of prediction results, which gave good results from the system created. The application has succeeded in identifying scam tokens with high accuracy. , increasing user security in crypto transactions.