Investigating the Use of Model-Based Method for Improving the Quality of Natural Language Requirements: A Controlled Empirical Study
MetadataShow full item record
Requirement engineering is a critical phase in software development that describes the customer needs and specifications for the software. Requirements are gathered through various sources and documented for a software product to be developed, written in Natural Language (NL). NL requirements are fault prone because they can be interpreted in different ways due its inherent imprecision, ambiguity, and vagueness. To address these problems, model-based requirements verification method called NLtoSTD (State Transition Diagram) is proposed. This paper evaluates the ability of NLtoSTD method in detecting faults when used on NL requirements and to improve its cognitive friendliness to the stakeholders. Motivated by the earlier study, we revised our proposed method and performed an empirical study. The participants employed the NLtoSTD method to inspect documents to identify ambiguities, incompleteness and inconsistencies. The experiment result shows an improvement over the previous results that the NLtoSTD is a method for verification of NL requirements.