IMPROVED SUPPORT VECTOR BASED RANKING MODEL AND FEATURE SELECTION METHODS FOR BUG REPORTS ANALYSIS

Abstract

Bug report tracking systems have been used toward make possible the maintenance and development of software. On the other hand, duplicate entries presented in the bug reports in such software system are able to significantly force productivity inside software project. This reduction in productivity happens since duplicate entries demand more time designed for search and examination of bug reports. When a new bug report is received, developers frequently require to reproducing the bug and performing code reviews to discover the cause, a process with the purpose of can be difficult and time consuming. To solve this problem in this paper, Support Vector Machine (SVM) system is proposed that considers features with the purpose of association related to lexical gap via the use of project precise Application Programming Interface (API) report to attach Natural Language parameters in the bug report by means of programming language with the purpose of build in the source code. The major contribution of this paper is to propose Support Vector Machine (SVM) ranking method which solves bug report problem related to source files with the purpose of permits the faultless integration of a different type of features. Support Vector Machine (SVM) ranking approach counts the frequency value toward each bug reports make use of formerly fixed bug reports as training examples designed for the proposed ranking-model in combination by means of a learning-to-rank technique; with the file dependency graph toward describe features with the purpose of confine a determine of code complexity. The extensive experimentation and comparisons is done to traditional ranking methods, it concludes that the proposed FOFR system of the impact with the purpose of features producing higher ranking accuracy and lesser code complexity.