This paper proposed a noble techniques Blizzard that automatically identifies buggy entities from source code using appropriate query reformulation and effective information retrieval.
Contributions of The Paper
Key contribution-
As bug report are used to build query for bug localization, it plays a vital role to locate the source code file. Although the existing studies used bug report as query to retrieve source code that contains buggy entities, the main challenges is that it could be noisy or it could lack of information. To overcome this barrier, this study classified the bug report in three classes based on the stack trace, program entities and natural language element in the bug reports. Later, they reformulate the query for efficient retrieval of source code.
Once the reformulated query is constructed, they submit it to Lucene for source code search (Lucene is widely adopted search engine which combines both boolean and VSM based searched methodologies). This study has been conducted on more than 5000 bug reports from six open source projects that has been written in Java language.
Blizzard outperformed BLUiR, AmalGam, BugLocator.
Besides this study investigate how Blizzard performs compared to the baseline when all results of the query are considered. Two types of baseline queries are considered in this case one is only the title of the bug reports and another one is (title + description) in the bug report.
Publisher
FSE
Link to The Paper
https://dl.acm.org/doi/10.1145/3236024.3236065
Name of The Authors
Mohammad Masudur Rahman , Chanchal K. Roy
Year of Publication
2018
Summary
This paper proposed a noble techniques Blizzard that automatically identifies buggy entities from source code using appropriate query reformulation and effective information retrieval.
Contributions of The Paper
Key contribution-
Comments
No response