Abstract
During the evolution of a software system, a large amount of information, which is not always directly related to the source code, is produced. Several researchers have provided evidence that the contents of mailing lists represent a valuable source of information: Through e-mails, developers discuss design decisions, ideas, known problems and bugs, etc. which are otherwise not to be found in the system. A technical challenge in this context is how to establish the missing link between free-form e-mails and the system artifacts they refer to. Although the range of approaches is vast, establishing their accuracy remains a problem, as there is no benchmark against which to compare their performance. To overcome this issue, we manually inspected a statistically significant number of e-mails pertaining to the ArgoUML system. Based on this benchmark, we present a variety of lightweight techniques to assign e-mails to software artifacts and measure their effectiveness in terms of precision and recall.