Title | Classifying Bug Reports into Bugs and Non-bugs Using LSTM |
Authors | Qin, Hanmin Sun, Xin |
Affiliation | Peking Univ, Beijing, Peoples R China. |
Keywords | bug classification LSTM |
Issue Date | 2018 |
Publisher | INTERNETWARE'18: PROCEEDINGS OF THE TENTH ASIA-PACIFIC SYMPOSIUM ON INTERNETWARE |
Citation | INTERNETWARE'18: PROCEEDINGS OF THE TENTH ASIA-PACIFIC SYMPOSIUM ON INTERNETWARE. 2018. |
Abstract | Studies have found that significant amount of bug reports are mis-classified between bugs and non-bugs, which inevitably affects relevant studies, e.g., bug prediction. Manually classifying bug reports helps reduce the noise but is often time-consuming. To ease the problem, we propose a bug classification method based on Long Short-Term Memory (LSTM), a typical recurrent neural network which is widely used in text classification tasks. Our method outperforms existing topic-based method and n-gram IDF-based method on four datasets from three popular JAVA open source projects. We believe our work can assist developers and researches to classify bug reports and identify misclassified bug reports. Datasets and scripts used in this work are provided on GitHub(1) for others to reproduce and further improve our study. |
URI | http://hdl.handle.net/20.500.11897/571649 |
DOI | 10.1145/3275219.3275239 |
Indexed | CPCI-S(ISTP) |
Appears in Collections: | 待认领 |