A software defect report assignment method based on self-attention mechanism

Abstract

The invention relates to a software defect report distribution method based on a self-attention mechanism. According to the defect report in the defect tracking system, the invention collects the defect report that has been allocated, and fully exploits the semantic information implied in the defect report by using the neural network and the attention mechanism, and calculates the matching degree between the developer and the target defect report to achieve Recommend the most suitable restorer. The invention solves the problem that the traditional defect report distribution technology relies on manual feature engineering work, utilizes the neural network to quickly and effectively extract and distribute information on the defect report, improves the efficiency of defect repair, and is suitable for the development and maintenance of large-scale software products.

Description

technical field [0001] The invention relates to a method for allocating software defect reports based on a neural network and an attention mechanism, in particular to a method for allocating software defect reports based on a self-attention mechanism. Background technique [0002] In large-scale open source software projects, such as Eclipse, Firefox, etc., as the scale increases and the version is updated, there will be a large number of defects. Managing and repairing software defects is a time-consuming, labor-intensive and difficult task. Software users and developers use defect tracking systems, such as Bugzilla, Bugfree, etc., to submit and record software defect reports. The quality management staff will estimate the severity of the submitted defect reports and assign fixers. Among them, according to the field and knowledge involved in the content of the defect report, the defect report is assigned to the appropriate developer, and this process is the personnel assig...

AI Technical Summary

Problems solved by technology

[0002] In large-scale open source software projects, such as Eclipse, Firefox, etc., as the scale increases and the version is updated, there will be a large number of defects. Managing and repairing software defects is a time-consuming, labor-intensive and difficult task

[0005] In the existing defect report allocation methods, a large number of manual feature engineering processes are often required, which requires rich domain knowledge and professional matching, and does not consider the semantic information of the defect report text, so it is difficult to effectively apply to actual software development and maintenance. process

Images