On the Influence of Models at Run-Time Traces in Dynamic Feature Location

Feature Location is one of the most important and common activities performed by developers during software maintenance and evolution. In prior work, we show that Dynamic Feature Location obtains better results working with models rather than source code. In this work, we analyze how the criteria to create the model traces influence the Dynamic Feature Location results. We distinguish between two different criteria: configuration and architecture. Our Dynamic Feature Location approach is composed of dynamic analysis, information retrieval at the model trace level, and information retrieval at the model level. The evaluation in a Smart Hotel tests whether the traces created following the two criteria modify the results of the Feature Location by measuring recall, precision, and the combination of both (F-measure). The results reveal that in 75% of the cases the traces that follow the architecture criterion outperform the traces that follow the configuration criterion.