Paper Review: On the Relationship between The Vocabulary of Bug Reports and Source Code

[usmimukherjee]

Publisher

ICSME

Link to The Paper

https://www.academia.edu/28272214/On_the_Relationship_between_the_Vocabulary_of_Bug_Reports_and_Source_Code

Name of The Authors

Sonia Haiduc, Wathsala Bandara

Year of Publication

2013

Summary

They conducted their study to find whether the bug reports and the related defect source code share important terms. They analyzed the vocabularies shared between the source code classes and the bug reports.

Contributions of The Paper

Their contributions are :

• To acquire evidence to support the implicit assumption that bug descriptions and the source code share some significant words which help to map the bug and the defect code.
• They proposed a technique to explore the common vocabularies obtained from the bug report and the patched classes. They analyzed these vocabularies to identify significant patterns and the existence of the relationship between source code and bug reports.

Observations and Findings :

• The number of unique terms increases with the bug report size.
• The larger the class, the more unique terms it contains
• Bug reports share more terms with the patched classes than the non patched classes in average
• The textual similarity between the bug reports and the patched classes is higher compared to non patched classes.
• The code location which are more verbose tent to contribute more to the common vocabulary between the patched classes and its relevant bug report.
• The names of the patched classes are more likely to have shared terms with relevant bug reports than the terms belonging to the other code locations
• The bug location is supported by a number of shared terms between the bug reports and the source code classes better than LSI does, but it does not perform better than Lucene.

Comments

No response