Generating Intelligent Summary Terms for Improving Knowledge Discovery in Software Bug Repositories

Software bug records are stored and managed using bug tracking tools. A software bug is characterized by a number of attributes like bug id, opened date, closed date, reported by, assigned to, summary (title), description and set of comments. Summary and description are the two important attributes of a bug. Description gives the detailed information about a bug, whereas summary (title) of a bug gives a quick glance and short information about a bug. The objective of this study is to discover the relationship between description and summary attributes of a bug and to find whether summary of a bug is really the compact and intelligent information of description of a bug. This finding helps in providing a new direction for faster knowledge discovery in a bug repository. Another objective of the work is to demonstrate that intelligent summary of a bug can be generated from description of bug using topic modeling techniques. In this work, topic modeling techniques are used to generate meaningful terms for framing the bug summary of software bugs which can be utilized for faster knowledge discovery. Topic modeling techniques can be utilized efficiently for generating intelligent summary from description of a software bugs and then the knowledge discovery can be performed using the intelligent summary only since it will reduce the volume of data for knowledge discovery. To demonstrate the presented approach, experiments are performed on three popular bug repositories namely, Android, Mozilla and MySql. Comparative analysis is carried using various performance parameters and in order to analyze the impact of present work, two knowledge discovery tasks namely, bug classification and duplicate bug identification are presented in this study.