Natural Language-Based Model-Checking Framework for Move Smart Contracts

Significant efforts have been dedicated to employing model-checking as a formal verification approach in the context of smart contracts. The utilization of these tools necessitates an in-depth knowledge on the part of the developer regarding both the programming language and the implementation of model-checking techniques. To provide accessibility to developers with basic language proficiency, we present a technique for developing a conversational application framework that can be seamlessly linked with any model-checking tool for the purpose of creating a smart contract. This architecture offers a robust and effective approach to the development of safe and dependable smart contracts. The utilization of natural language processing techniques in conjunction with neural networks is employed for this objective. Using this methodology, a prototype implementation for Move smart contracts has been created and is used with the VeriMove model-checking tool. Using the offered graphical user interface, we were able to successfully build, compile and test Move smart contracts across four different classes of smart contracts. This strategy effectively decreases the amount of time and effort needed for manual coding and debugging. In addition, the use of the VeriMove model-checking tool guarantees that the smart contracts produced are devoid of any potential vulnerabilities and flaws.