Evaluation of natural language requirements in an industrial case study

Automatic evaluation of natural language requirements documents has been proposed as a means to improve the quality of the system under development. QuARS (Quality Analyzer for Requirements Specifications) was introduced as an automatic analyzer of such requirement documents, developed by ISTI-CNR. In this paper we show how this tool has been applied to a large collection of requirements produced inside the EU/IP MODTRAIN project. In this project several industrial partners have contributed to the collection by proposing requirements coming from their own products, with the purpose to define an European standard for the Train Control and Monitoring Systems (TCMS). To better fit the project needs, QuARS has been extended to be interfaced with the requirement management tool adopted in the project, and adapted to specific and domain dependent aspects of the MODCONTROL requirements. In particular the tool is now able to handle a more complex and structured data format containing metadata. The calculation of readability metrics and other statistical indexes has also been improved.