Bilgi geri getirimi ve makine öğrenmesi algoritmalarını kullanarak yazılımda hata konumlandırılması

Software quality assurance is crucial for the success of software. In large scale software projects, bug localization is a difficult and costly process. Many issues or bugs may be reported at both the development and maintenance phases of the software development lifecycle. Bug fixing has an essential role in software quality assurance, and bug localization is the first step of this process. Bug localization (BL) is time-consuming since the developers should understand the flow, coding structure, and logic of the program. Hence, it is crucial for developers to discover the location of the bug. In general, source codes and bug reports are used for identifying bug location with the help of many different techniques. Information retrieval-based bug localization (IRBL) also uses the information of bug reports and source code to locate the section of code in which the bug occurs. It is not possible to apply state-of-the-art approaches having a satisfactory performance to new projects according to the diversity of software development languages, design patterns and development standards. This thesis proposes a novel algorithm, Adaptive Attribute Weighting (AAW), to adopt a new software project for BL processes. This thesis also includes the development of a new BL tool, BugSTAiR, in order to work on with all software projects. Experimental studies demonstrate the capability of the AAW algorithm and also the BugSTAiR tool on both real-life and experimental datasets, including commercial software projects which are developed with different languages and improvement in performance compared to the existing algorithms.