A Quality Evaluation Process for Collaborative Requirements: An Industrial Case Study

When requirements come from different companies, as it often happens in big industrial projects, it is crucial for the success of the project to offer means and automatic tools for the early validation of the requirements. This paper introduces an approach for quality evaluation of natural language (NL) requirements documents performed with the help of an automatic support tool. The key factors of the approach are the concepts of readability, defects and errors detection in NL requirements documents (RDs) with support of the tools QuARS that exploits for the analysis NL processing techniques. The approach has been successfully applied to the collection of requirements for a Train Control and Monitoring System (TCMS), produced by writing several requirements coming from different people. The application was carried out inside the EU/IP MODTRAIN project, subproject MODCONTROL on the MODTRAIN requirements database containing more than 5500 requirements.