Robustness code summary generation method based on self-attention mechanism

Abstract

The invention discloses a robustness code summary generation method based on a self-attention mechanism. The method comprises the steps that firstly, high-quality codes and description corpus pairs (query description texts and reply code texts) in a programming community are extracted; then redundant information of the codes and description corpus pairs is filtered out; then the query descriptiontexts corresponding to the codes are converted into declarative statements; lastly, a code summary of a sequential model based on the self-attention mechanism is generated. The robustness code summarygeneration method based on the self-attention mechanism has the advantages that the redundant information and noise content can be effectively removed, the automatic evaluation and artificial evaluation accuracy of the generated summary are both improved, and the evaluation result is superior to that of an existing baseline method.

Description

technical field [0001] The invention belongs to the cross field of software engineering and natural language processing technology, and specifically relates to a method for generating robust code summaries based on the two advantages of wavelet time-frequency transform EM algorithm and sequence model based on self-attention mechanism, especially suitable for Code description data with noisy information in the programming community. Background technique [0002] In the evolution of large-scale software projects, code comments are the key work in software maintenance. High-quality comments often provide efficient reference information for developers to understand and update code. Although there are many tools and methods for program understanding, it still takes longer to understand the code than to write it. At the same time, developers tend to skim the code (code entities such as method signatures) to understand the program to reduce the workload, but this ignores key infor...

AI Technical Summary

Problems solved by technology

[0010] The present invention aims to improve the accuracy and naturalness of the code summary generation effect, and has strong robustness when processing the code-description (descriptive text of the question-the code fragment of the reply) noise corpus pair extracted from the programming question-and-answer community. It can overcome the noise introduced by directly collecting parallel corpus to train the summary generation model, and introduces the self-attention self-attention mechanism in the sequence model, which can reduce the dependence on long-distance sequences in the coding structure when performing seq2seq sequence learning, resulting in Issues with poor summary generation

Images