A Controlled Language to Assist Conversion of Use Case Descriptions into Concept Lattices

Abstract

System requirements are often incomplete and riddled with contradictions leading to failed projects. Formal methods have been offered to minimise these problems. However the use of such techniques requires highly trained specialists and results in a process that users have little involvement with. We have developed a viewpoint development approach, known as RECOCASE, to capture requirements from multiple viewpoints directly from the users which are automatically modelled to identify and reconcile differences between stakeholder requirements. The requirements are captured as use case descriptions in natural language according to current popular practice in object-oriented system development. We use LinkGrammar on the use case sentences to output flat logical forms which we translate into crosstables and generate concept lattices using Formal Concept Analysis. To improve the output of our natural language process we have designed a controlled language to constrain the grammar and style. In this paper we introduce our natural language approach and describe and justify a controlled language we have developed to assist natural language translation.