An annotation assisted smart contracts generation method

Abstract

Abstract With the rapid development of blockchain technology, the demand and complexity of smart contracts have sharply increased. However, smart contracts not only have high security requirements, but also have unique development languages that conventional software developers find difficult to quickly adapt to. Therefore, how to efficiently develop secure and reliable smart contracts is a key issue. Therefore, we propose an annotation guided intelligent contract automatic generation method based on the Char-RNN network. It utilizes annotation information from the source code of smart contracts as semantic assist, and enhances the clustering performance for functionally similar smart contracts, which can obtain more accurate features of smart contracts. At the same time, to enhance the applicability of the model, the method achieves automatic generation of code at multiple levels such as contracts, functions, interfaces, libraries, etc., to meet the development needs of different users. To evaluate the effectiveness of our method, we conducted experiments on the automatically generated smart contract using the BLUE metrics and VaaS security detection tool. The experimental results showed that compared with existing methods, the smart contract generated by our method improved the average BLEU score by 27% and the accuracy by 7.6%.