A Method for Automatic Classification of Software Defects Based on Association Rules

Abstract

The invention discloses a method for automatically classifying software defects based on association rules, which comprises the following steps: first extracting the text content in the defect report, and preprocessing it; then randomly sampling the defects whose state is VERIFIED FIXED, and constructing defects to be classified Then according to the defect appearance and defect occurrence cause, each defect in the to-be-classified defect set is labeled with the defect appearance category and the defect occurrence cause category to obtain the defect appearance category classification set and the defect occurrence cause category classification set; and then excavate different defect appearance categories, The association rules between defect occurrence cause categories are selected, and strong association rules are selected from them and converted into a relationship matrix; finally, the deep learning method is used to train the classification set of defect appearance categories to obtain an automatic defect classifier. The present invention simultaneously performs fine-grained automatic classification of defects from two dimensions of defect appearance and cause, which solves the shortcoming of inability to accurately classify defects due to a small amount of defect information in the prior art.

Description

technical field [0001] The invention belongs to the field of software maintenance, in particular to an automatic software defect classification method based on association rules. Background technique [0002] Software defects are one of the most serious problems to be solved in the process of software quality maintenance. With the continuous expansion of software scale, a large number of new defects are introduced, and intelligent defect repair has become the focus of industry research. To fix defects in a timely and effective manner, you first need to identify the appearance of the defect and the cause of the defect. However, the defect reports submitted by users are often incomplete and inaccurate. They only include the steps to reproduce the defect, the actual results and the expected results. Developers can relatively easily However, it takes a lot of time to analyze the cause of the defect. Accurate defect classification is an important means to improve the efficiency ...

AI Technical Summary

Problems solved by technology

However, none of the above studies have explored the relationship between fine-grained defect appearances and cause categories, nor have they applied these relationships to automatic defect classification.

In addition, there are some works dedicated to the research of automatic defect classification technology, but most of these technologies use artificial feature engineering and shallow neural network classification models, and artificial feature engineering does not have strong versatility, and often needs to be combined with the understanding of different software. Choose different features

In addition, the text context is ignored in the text representation, each word is independent of each other, and semantic information cannot be represented

Images