Mining Effort Data from the OSS Repository of Developer's Bug Fix Activity

Syed Nadeem Ahsan1, Muhammad Tanvir Afzal2, Safdar Zaman1, Christian Gütel2,3, Franz Wotawa1

1Institute for Software Technology (IST) Graz University of Technology, Inffeldgasse 16b/II 8010, Graz, Austria

2Centre for Distributed and Semantic Computing (CDSC), Mohammad Ali Jinnah University, Islamabad, Pakistan

3School of Information Systems (SIS), Curtin University of Technology, Perth, Western Australia


This paper appears in: Journal of IT in Asia, Vol 3 (2010), pg: 67-80
Issue Date: Nov 2010
Print ISSN: 1823-5042

Abstract:

During the evolution of any software, efforts are made to fix bugs or to add new features in software. In software engineering, previous history of effort data is required to build an effort estimation model, which estimates the cost and complexity of any software. Therefore, the role of effort data is indispensable to build state-of-the-art effort estimation models. Most of the Open Source Software does not maintain any effort related information. Consequently there is no state-of-the-art effort estimation model for Open Source Software, whereas most of the existing effort models are for commercial software. In this paper we present an approach to build an effort estimation model for Open Source Software. For this purpose we suggest to mine effort data from the history of the developer’s bug fix activities. Our approach determines the actual time spend to fix a bug, and considers it as an estimated effort. Initially, we use the developer’s bug-fix-activity data to construct the developer’s activity log-book. The log-book is used to store the actual time elapsed to fix a bug. Subsequently, the log-book information is used to mine the bug fix effort data. Furthermore, the developer’s bug fix activity data is used to define three different measures for the developer’s contribution or expertise level. Finally, we used the bug-fix-activity data to visualize the developer’s collaborations and the involved source files. In order to perform an experiment we selected the Mozilla open source project and downloaded 93,607 bug reports from the Mozilla project bug tracking system i.e., Bugzilla. We also downloaded the available CVS-log data from the Mozilla project repository. In this study we reveal that in case of Mozilla only 4.9% developers have been involved in fixing 71.5% of the reported bugs.

Keywords:

Software repository, mining effort data, estimation models, developer expertise and open source software.

 

 

 

©Copyright of  Universiti Malaysia Sarawak, 94300 Kota Samarahan, Sarawak, Malaysia http://www.unimas.my