An IR-Based Artificial Bee Colony Approach for Traceability Link Recovery

Multiple software development activities, including change impact analysis, requirements validation, maintenance and evolution of software benefit from adequate requirements traceability link recovery practices. Requirements traceability link recovery increases the overall quality of software products; however, companies often are unsuccessful implementing them due to lack of communication and strict deadlines. Several approaches for semi-automatic link recovery across requirements and source code in which textual analysis and information retrieval (IR) techniques are the baseline have been developed, but there is a need for methods that further enable the automatic generation of links. To advance automatization in the link recovery process, we investigate it as a combinational problem using an optimization approach. In this paper, we studied the requirements traceability recovery problem as a big search space formed by multiple pairs (requirements and source code classes), where requirements are matched to code elements. The artificial bee colony (ABC) algorithm is adapted for searching for the solution that maximizes an objective function which is calculated by a weighted cosine similarity where weighs for each term in the textual content of requirements and source code are defined according to the term frequency (TF-idf). Evaluation over three data sets demonstrates the effectiveness of the proposed approach. It returned high precision and recall values for the recovered links.