Generating Logical Representations for Natural Language Requirements Using Syntactic Dependencies and Norm Analysis Patterns

摘要

Requirements expressed in Natural Language are often ambiguous, inconsistent and, not amenable to automated analysis and validation. Formal approaches like mathematical or logical formalism to requirements representation offer possible solution to these problems. However, formal specifications are not widespread in industry as analysts and business users find them difficult to comprehend. In this paper, we present an approach to translate Natural Language representation of requirements to Logical representations. We have used Courteous logic, a non-monotonic form of logic, for the purpose. Our approach is based on syntactic dependency analysis of requirements statements, Norm Analysis Patterns and Grammatical Knowledge Patterns. The analyzed information is stored in frame-based structured representation for the requirements. These structured representations are translated to courteous logic form. We also report the effectiveness of our approach through the case-studies conducted.