A Case Study in Systematic Improvement of Language for Requirements

Abstract

The challenges to requirements from linguistic factors are well-known. This work concerns an approach to communicating requirements with greater fidelity among stakeholders through accommodation of cognitive habits and limits. To instantiate this approach, we synthesized linguistic principles into a method to generate high-quality representations of domain concepts to form the base of a project lexicon. The representations are further organized into a knowledge base that records relationships of interest. We hypothesize that the method leads to representations free of certain faults that compromise communicative fidelity. To investigate, we executed a case study in which the method was applied to the domain semantics of a medical device. Our representations compared favorably to pre-existing versions; further, analysis of the artifact as a whole supported new assertions about the application domain. These results indicate that particular language issues can be systematically managed, and that new information is available in the process.

Research Directions in Requirements Engineering

“…Requirements analysis assesses the quality of requirements models and documentation. Most of the research in this area focuses on new or improved automated techniques for detecting errors in models, where an "error" can be ambiguity [56,81,129,153], inconsistency [55,70,107], an unknown interaction among requirements [23,67,125], a possible obstacle to requirements satisfaction [96,152], or missing assumptions [11].…”

Regulatory Requirements Traceability and Analysis Using Semi-formal Specifications

“…In Section 4, we show how the LRSL can be used to shape conditionality of regulatory coverage, which is enabled by the tool-supported ability to trace definitions across a single regulation, or across multiple regulations as definitions are shared across laws. Reusing technical terminology improves requirements engineering practices, as it avoids misconceptions among stakeholders and competing viewpoints that introduce inconsistency into design specifications [33]. Zave and Jackson have noted the importance of grounding terminology in the reality of the environment to which a machine will be built [37].…”

Semantic parameterization

“…In this article, we focus on the process to model natural language descriptions of systems and their environments. Therefore, we first review the role of ontology in modeling domain knowledge before reviewing two popular methods for acquiring formal specifications from natural language descriptions: controlled languages [Cregan et al 2007;Konrad and Cheng 2005;Kaljurand and Fuchs 2007;Smith et al 2002] and standard lexicons [Cysneiros and Leite 2004;Kaindl 1996;Overmyer et al 2001;Wasson et al 2003;Wasson 2006]. We conclude by reviewing types of ambiguities identified by Gause and Weinberg [1989].…”

“…Several researchers have recognized the need to better align natural language requirements and formal models to prevent the loss of original context [Potts 1997]; incorporate knowledge about the system environment [Jackson 1997;van Lamsweerde 2000;Mylopoulos et al 1997;Nuseibeh and Easterbrook 2000]; improve traceability [Ramesh and Jarke 2001]; apply formal analysis to requirements concepts [van Lamsweerde 2000;Nuseibeh and Easterbrook 2000]; and reduce ambiguous terminology [Gause and Weinberg 1989;Jackson 1997;Denger et al 2003;Wasson et al 2003;Wasson 2006]. In this article, we focus on the process to model natural language descriptions of systems and their environments.…”

“…Similarly, Kaindl [1996] showed how to identify binary relationships between nouns in natural language definitions and map them to new classes. Wasson et al employed a domain map to facilitate effective communication between domain experts and engineers [Wasson et al 2003;Wasson 2006]. The domain map contains technical words classified into hierarchies of superordinate and subordinate terms and is used to identify ambiguous terms based on their commonality and domain-specific interpretation.…”