Open AccessAssessment of Software Bug Complexity and Severity using Evolutionary SOM Scheme
Author(s) -
Malvika Rao,
AUTHOR_ID,
Dr.P Suryanarayana Babu,
AUTHOR_ID
Publication year - 2019
Publication title -
international journal of engineering and advanced technology
Language(s) - English
Resource type - Journals
ISSN - 2249-8958
DOI - 10.35940/ijeat.f9257.088619
Subject(s) - software bug , computer science , software regression , software , software quality , software development , process (computing) , software maintenance , reliability engineering , software construction , software metric , software engineering , engineering , operating system
The software defect prediction and assessment plays a significant role in the software development process. Predicting software defects in the earlier stages will increases the software quality, reliability and efficiency, the cost of detecting and eliminating software defects have been the most expensive task during both development and maintenance process, as software demands increase and delivery of the software span decreased, ensuring software quality becomes a challenge. However, due to inadequate testing, no software can pretend to be free from errors. Bug repositories are used for storing and managing bugs in software projects. A bug in the repositories is recorded as a bug report. When a bug is found by a tester its available information is entered in defect tracking systems. During its resolution process a bug enters into various bug states. These defect tracking systems enable user to give the information about the bugs while running the software. However, the severity prediction has recently gained a lot of attention in software maintenance. Bugs with greater severity should be resolved before bugs with lower severity. In this paper an evolutionary interactive scheme to evaluate bug reports and assess the severity is proposed. This paper presents a Software Bug Complexity Cluster (SBCC) using Self Organizing Maps. In this SBCC a feature matrix is built using bug durations and the complexities of software bugs are categorized into distinct clusters including Blocker, Critical, Major, Trivial and Minor by specifying negative impact of the defect using two different techniques, namely k-means and SOM. Bug duration, proximity error and pre-defined distance functions are used to estimate the accuracy of different bug complexities. Our systematic study found that SOM's proximity error and fitness have greater performance and efficiency than K-Means. The collected results showed better performance for the SBCC with respect to fitness and cluster proximity error.