A Survey of Software Defects Research Based on Deep Learning

Abstract

In the process of software development, software defects are inevitable. How to quickly and accurately identify defects and accurately deliver bug reports to the most appropriate repair personnel is very important for defect repair. In recent years, with the development of artificial intelligence, deep learning has been widely used in software defect research. This paper summarizes and analyzes the progress of software defect research based on deep learning in the past three years from four perspectives of software defect prediction, software defect identification, software defect analysis and bug report assignment. They are software defect prediction technology based on the TextCNN model and TextRNN model, a software defect identification technology based on LSTM, BiLSTM, DNCC, a software defect analysis technology based on the DAKSM model, and a software bug report assignment technology based on Atten-CRNN model. And compared these mentioned deep learning-based techniques with previous techniques in the field. After experiments, it is concluded that these models have good accuracy. At the same time, the related technologies involved are introduced in detail, as the problems that may be encountered in future research in this field and the development prospects prospect.