CORG: a Component-oriented synthetic textual requirements generator

The majority of requirements formalisation techniques operate on textual requirements as input. To establish and verify the reliability and coverage of such techniques, a large set of textual requirements with diverse structures and formats is required. However, such techniques are typically evaluated on only a few manually curated requirements that do not provide enough coverage of the targeted structures. Motivated by this problem, we introduce a Component-oriented synthetic textual requirements generator (CORG) that can generate large numbers of synthesised diverse-structure textual requirements, along with key components breakdowns. CORG utilises a controlled random-selection (CRS) strategy throughout the backtracking-based generation. We evaluate the coverage, diversity, performance and correctness of CORG. The evaluation results show that CORG can generate comprehensive diverse-structure combinations in reasonable time without being affected by the size of the produced requirements.