Improving predictions about bug severity by utilizing bugs classified as normal

Kwanghue Jin, Amarmend Dashbalbar, Geunseok Yang, Byungjeong Lee, Jung Won Lee

Research output: Contribution to journalArticlepeer-review

10 Scopus citations

Abstract

More bugs have been generated because today's software has become large and complex. Some of these bugs are critical, while others are trivial. Because accurate prediction of bug severity enables software developers to effectively solve software problems, that accuracy aids software development and project planning. Therefore, this study presents a reliable approach to improve predictions about bug severity by utilizing bugs classified as normal, which is the default level specified in a submitted report. This approach uses attributes as well as text information in bug reports to produce more accurate prediction results. Bug reports in open source projects such as Mozilla and Eclipse were used in the experiments. The result shows that this approach performs better than other studies.

Original languageEnglish
Pages (from-to)933-942
Number of pages10
JournalContemporary Engineering Sciences
Volume9
Issue number19
DOIs
StatePublished - 2016

Keywords

  • Bug Severity Prediction
  • Classification
  • Machine Learning
  • Normal Bugs
  • Project Planning
  • Software Development

Fingerprint

Dive into the research topics of 'Improving predictions about bug severity by utilizing bugs classified as normal'. Together they form a unique fingerprint.

Cite this