Paper 2020/574

Towards Interpreting Smart Contract against Contract Fraud: A Practical and Automatic Realization

Ming Li, Jian Weng, Anjia Yang, Jiasi Weng, and Yue Zhang

Abstract

With the advent of Ethereum blockchain, a new type of contract, named smart contract, is emerging nowadays, enabling people to describe complicated logics as automatically executable computer programs. Due to the lack of the computer background and special design of smart contacts, many people may have difficulty in understanding smart contracts, while they still have urgent demands to leverage them to build more trusted environment with others for the obvious advantages of blockchain. However, there does not exist an automatic technique to guide those people who do not have much background on smart contracts. Furthermore, a new wave of smart contracts fraud has been brought to them and caused serious economic loss. To address this challenge, we investigate the general rules of the smart contract codes and propose a new approach, called SMTranslator, to enable people without computer background to understand and operate Ethereum smart contracts. Particularly, we first translate smart contracts into standard structured files and identify core statements for each function based on principles of dependency weights. By exploiting the custom natural language generation, we then describe the documents that can provide correct and understandable descriptions. The visibility and vulnerability graph are also generate to alert people against the fraud issue. Furthermore, to conduct the experiments, we collect numerous smart contracts from Ethereum and select 60 volunteers. Extensive experimental results demonstrate that SMTranslator can automatically interpret smart contracts and most of the generated documents can be understood and guide volunteers to utilize smart contracts, which shows the feasibility and effectiveness of our approach.