Univ.-Prof. Dipl.-Ing. Dr. Martin Pinzger | Mi, 26.03.2014 | 17:00-18:30, HS 2

Abstract: Software repositories store a wealth of information about software projects including data about failures that get reported by testers and users. One of the main themes of the mining software repositories research is to use this rich information to prevent failures, for instance by training models that check the current release of a system and point out potential bugs. In this talk, I present examples of my research in mining software repositories to identify failure-prone source files and methods in several open source projects and failure-prone binaries in the Microsoft Windows Vista project. While the prediction models show promising results in pointing out failure-prone entities they leave room for many interpretations on what to do in order to prevent failures. Based on these examples and my observations I discuss several challenges of mining software repositories and potential applications.

This talk is part of / dieser Vortrag ist Teil der Ringvorlesung Informatik und Informationstechnik SS2014.