Abstract
Software developers are interested in requirement traceability to e.g., verify if all requirements are covered by a system design specification. Based on the assumption that related artifacts contain related terms, researchers have developed, used, and extended algorithms that identify related terms and subsequently infer which artifacts are related (i.e., there is a traceability link between them). Source code is not as verbose as a natural language description, which reduces the applicability of algorithms that precisely rely on such a commonality. This paper extends the Vector Space Model using tf*idf term weights to improve the identification of traceability links between source code and requirements. To this extent, we modify the way how requirements are identified and to include user feedback. We show that the inclusion of user feedback significantly improved the number of correctly identified requirements. © 2012 by Knowledge Systems Institute Graduate School.