Introducing Collaboration for Locating Features in Models: Approach and Industrial Evaluation

Abstract

Feature Location (FL) is one of the most important tasks in software maintenance and evolution. However, current works on FL neglected the collaboration of different domain experts. This collaboration is especially important in long-living industrial domains where a single domain expert may lack the required knowledge to fully locate a feature, so the collaboration among different domain experts could alleviate this lack of knowledge. In this work, we address collaboration among different domain experts by automatically reformulating their feature descriptions. With our approach, we extend existing FL approaches based on Information Retrieval and Linguistic rules to locate features in models. We evaluate our approach in a real-world case study from our industrial partner, which is a worldwide leader in train manufacturing. We analyze the impact of our approach in terms of recall, precision, and F-Measure. Moreover, we perform a statistical analysis to show that the impact of the results is significant. Our results show that our approach for collaboration boosts the quality of the results of FL.