Schlutter, AaronVogelsang, Andreas2018-12-032018-12-0420181613-0073https://depositonce.tu-berlin.de/handle/11303/8642http://dx.doi.org/10.14279/depositonce-7776Complex systems such as automotive software systems are usually broken down into subsystems that are specified and developed in isolation and afterwards integrated to provide the functionality of the desired system. This results in a large number of requirements documents for each subsystem written by different people and in different departments. Requirements engineers are challenged by comprehending the concepts mentioned in a requirement because coherent information is spread over several requirements documents. In this paper, we describe a natural language processing pipeline that we developed to transform a set of heterogeneous natural language requirements into a knowledge representation graph. The graph provides an orthogonal view onto the concepts and relations written in the requirements. We provide a first validation of the approach by applying it to two requirements documents including more than 7,000 requirements from industrial systems. We conclude the paper by stating open challenges and potential application of the knowledge representation graph.en004 Datenverarbeitung; Informatiknatural language processingknowledge representationautomotive softwarerequirementsindustrial systemsStanford CoreNLPSemantic Role LabelingNLPConference Objecturn:nbn:de:0074-2075-4