A pattern matching and clustering based approach for supporting requirements transformation

One problem in requirements transformation from informal prose requirements to formal specifications is natural language processing. Although natural language understanding is still not practical for general requirements processing, syntactical information is useful for concept extraction. This paper discusses the correspondence between requirements sentence structure patterns and the events/transition concepts. It addresses how this correspondence might be utilized to extract events and transitions from conditional sentences. A pattern matching and clustering based approach is proposed to support event and transition extraction. The approach first selects requirements that are likely to contain event/transition information and then groups similar requirements to extract events. Once events are identified, the original requirements are clustered based on events to identify, related transition information. The approach is simple and does not require a parser