A code reviewer recommendation method based on history

Abstract

The present invention provides a method for recommending code reviewers based on historical records. The method utilizes the feedback information on codes in historical review records and integrates time factors to sequentially determine the professional weights of different reviewers for a piece of code at different times. Path similarity analysis to finally determine the reviewer's recommendation sequence; this method provides a robust and efficient code reviewer recommendation method, it first extracts the feedback from different reviewers in the historical records for a certain piece of code, the number of comprehensive feedback, Date, frequency and other information to determine the professionalism of different reviewers for a piece of code, and then use the string comparison method based on path similarity, combined with time factors, and use the principle of communication to build a path similarity model based on weight, and finally get different The reviewer's priority order for a piece of code.

Description

Technical field[0001]The present invention relates to the field of code review, and more specifically, to a modern code reviewer recommendation method based on historical records.Background technique[0002]With the further development of the software engineering field, the number of codes and the difficulty of maintenance have also increased. Modern code audits for the purpose of reducing maintenance costs and improving code quality and efficiency have received more and more attention. However, from among the many candidate reviewers, finding the most suitable reviewer for a piece of code to be reviewed has increasingly become a bottleneck restricting the further development of modern code review. The current existing reviewer recommendation methods are mainly divided into two types: a method based on path similarity and a method based on code line review records. The reviewer recommendation method based on path similarity is to analyze the path similarity between the path of the cod...

AI Technical Summary

Problems solved by technology

[0003] Among them, in the prior art, the code reviewer recommendation method based on path similarity does not consider the professionalism level of different reviewers for the same code in different time periods, and the method regards the contributions of reviewers participating in the review as consistent; At the same time, in terms of measuring professionalism, this method only considers the path similarity, ignoring the importance of other feedback information, resulting in low accuracy of the algorithm as a whole; the reviewer recommendation method based on code line review records focuses on code The modification audit information of a line is based on the code as the basic unit. However, in actual production, the situation of modifying the same line of code is relatively small. For systems that lack line modification audit information, this method shows great uncertainty.

Images