On system sequence descriptions

Context: Use cases (UCs) are widely used to specify the functionality of a SW system. A UC is usually worked out in a Main Success Scenario and several other (Alternative) Scenarios. To make the overall structure of the UC clear and to prepare for the software implementing the UC, these scenarios must be integrated into one structure per UC. Question/problem: How to integrate the different scenarios of a UC into one structure? Our solution: We propose so-called (textual) System Sequence Descriptions (SSDs). We introduce a suitable (context-free) grammar for our SSDs. We can express all usual constructs with our SSDs, such as primary and secondary actors, basic steps, the internal responsibilities of the system, sequential composition, arbitrary order, loops/repetition, conditionals, alternatives, options, choices, definitions, and calls/'Includes'. To support validation of the resulting SSDs (with integrated scenarios) and check it with the users (Requirements quality assessment), we give (inductive) translation rules to translate the SSDs to natural language (Natural Language Generation for RE). Additionally, we give (inductive) rules to generate graphical SSDs (like the more familiar UML-diagrams) from our textual SSDs. This might support validation too. Results: With this new artefact design we can easily integrate the different scenarios of a UC into one clear SSD and also check the result with the users. That integrated SSD clarifies the overall structure of the UC and forms a suitable basis for implementation. With a nontrivial example we illustrate that the proposed grammar is very practical and that the approach scales up easily. Main contribution: The paper describes novel technical solutions for the application of NL-technologies to RE-relevant artefacts. The paper includes additional solutions to the RE-problem of validation.