Method for automatically generating Java code annotation based on Transformer and mixed code representation

Abstract

The invention provides a method for automatically generating a Java code annotation based on Transformer and mixed code representation, which comprises the following steps of: S1, downloading a Java project and constructing a code library; S2, in a serialization processing layer, converting into a code token vector and an SBT vector based on AST traversal; S3, in the coding layer, using a Code coder and an SBT coder, enabling the Code coder to extract lexical information from the source code, and using the SBT coder for obtaining structural information of the code; and S4, in a decoding layer,decoding the semantic information to generate annotations. The method has the beneficial effects that the method is used for generating code annotations, particularly encoding Codes and AST-based SBTtraversal sequences on an encoding layer, and combining semantic information learned by the Codes and the AST-based SBT traversal sequences to capture semantic information of source codes.

Description

technical field [0001] The invention relates to the technical field of computer applications, in particular to a method for automatically generating Java code annotations based on Transformer and mixed code representation. Background technique [0002] In the process of software development and maintenance, the comments corresponding to the code often have problems such as missing, insufficient or mismatching with the actual content of the code, but manually writing code comments is time-consuming and laborious for developers, and the quality of comments is difficult to guarantee, so it is urgent Researchers need to come up with effective methods for automatically generating code comments. [0003] Code comment generation aims to generate a natural language description of the source code, which can help developers understand the program, thereby reducing the time cost of software maintenance. Recently, most state-of-the-art techniques utilize Seq2Seq models based on RNN (Re...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method for automatically generating Java code annotations based on Transformer and mixed code representation, to solve the problem of poor program readability and poor comprehensibility caused by the lack of code annotations in the software development and maintenance process in the prior art , The problem of increased software development and maintenance costs, this method is used for code comment generation, especially at the encoding layer to encode Code and AST-based SBT traversal sequences, and combine the semantic information learned by the two to capture the semantics of the source code Information; the present invention realizes the automation of code annotation generation, generates concise and accurate annotations for codes, improves the readability and comprehensibility of codes, reduces code development and maintenance costs, and improves code development and maintenance efficiency

Images