Visual and User-Defined Smart Contract Designing System Based on Automatic Coding

Smart contract applications based on Ethereum blockchain have been widely used in many fields. They are developed by professional developers using specialized programming languages like solidity. It requires high requirements on knowledge of the specialized field and the proficiency in contract programming. Thus, it is hard for normal users to design a usable smart contract based on their own demands. Most current studies about smart contracts focus on the security of coding while lack of friendly tools for users to design the specialized templates of contracts coding. This paper provides a visual and user-defined smart contract designing systems. It makes the development of domain-specific smart contracts simpler and visualization for contract users. The system implements the domain-specific features extraction about the crawled data sets of smart contract programs by TF-IDF and K-means++ clustering algorithm. Then, it achieves the automatic generation of unified basic function codes by Char-RNN (improved by LSTM) based on the domain-specific features. The system adopts Google Blockly and links the generated codes with UI controls. Finally, it provides a set of specialized templates of basic functions for users to design smart contracts by the friendly interface. It reduces the difficulty and costs of contract programming. The paper offers a case study to design contracts by users. The designed contracts were validated on the existing system to implement the food trading and traders' credit evaluation. The experimental results show that the designed smart contracts achieve good integration with the existing system and they can be deployed and compiled successfully.