A natural language processing approach for avoidance of feature interactions

This paper proposes an approach to detect semantic anomalies in requirements written in natural language in order to avoid features interactions. This approach is based on the underlying idea that semantic anomalies in requirements may produce feature interactions in other phases of the service creation process. The paper focuses on a methodology to build from the requirements in natural language a conceptual representation on which inferences will be executed in order to detect anomalies. The methodology includes also a linguistic study of anomalies and a linguistic modeling of the notions of the domain. Some examples are presented to show that semantic anomalies correspond to potential feature interactions.