Java function annotation automatic generation method based on program analysis

Abstract

The invention relates to a Java function annotation automatic generation method based on program analysis, which comprises the following steps of: firstly, collecting a large number of open source projects on GitHub, and extracting Java function information and a calling relationship based on a program analysis technology to form a training data set; then training a translation model from a function to annotation based on a recurrent neural network technology in the deep learning field, automatically generating annotation for the function, and the effect is obvious; the problems of scarcity, non-standardization, inconsistency and the like of existing function annotations in a current project are effectively solved, and code readability and maintainability are improved.

Description

technical field [0001] The invention belongs to the field of computer technology, especially the field of software technology. The invention proposes a method for automatically generating annotations of Java functions based on program analysis, which can automatically generate annotations for functions by constructing a recurrent neural network model. It effectively solves the scarcity, irregularity, and inconsistency of existing function annotations in the current project, thereby improving code readability and maintainability. Background technique [0002] With the rapid development of Internet technology and the rapid changes in the software industry, more and more projects have begun to use the distributed collaborative development model. Developers usually create their own branches, develop on their own branches, and then merge the branches into the main branch. In this way, the characteristics of distributed development can be utilized to the greatest extent, and dev...

AI Technical Summary

Problems solved by technology

At the same time, common problems in the process of distributed collaborative development are: when developers develop or review code, they often spend more time understanding the code written by other developers

However, the previous work did not make full use of the information contained in the project that is beneficial to annotation generation and the advantages of deep learning, resulting in certain defects in the accuracy, comprehensibility and fluency of the generated annotations.

Images