In EDS ansehen

Requirements specification using templates: a model-driven approach.

Gespeichert in:
Bibliographische Detailangaben
Titel: Requirements specification using templates: a model-driven approach.
Autoren: Darif, Ikram1 (AUTHOR) ikram.darif.1@ens.etsmtl.ca, El Boussaidi, Ghizlane1 (AUTHOR) ghizlane.elboussaidi@etsmtl.ca, Kpodjedo, Sègla1 (AUTHOR)
Quelle: Software & Systems Modeling. Dec2025, Vol. 24 Issue 6, p1897-1934. 38p.
Schlagwörter: *REQUIREMENTS engineering, *MODEL-driven software architecture, *NEEDS assessment, *QUALITY assurance, *NATURAL language processing, *DESIGN templates, *CONCEPTUAL models
Abstract: Requirements specification and verification are crucial processes of software development. These processes are particularly costly for safety critical systems due to the high number of requirements and their complexity. For such systems, it is important to use natural language for the specification, as requirements need to be readable by non-technical stakeholders and certification agents. To mitigate the inherent ambiguity caused by the use of natural language, controlled natural languages (CNL) are introduced as a means to constrain the specification while maintaining readability. In this paper, we leverage model-driven engineering (MDE) to propose RESPECT, REquirements SPECification using Templates, a CNL-based approach for requirements specification and verification. The fundamental idea of RESPECT is to use MDE techniques to: 1) model requirements' templates and thus ease their creation, implementation and evolution and 2) link the template models to existing domain models to support, to some extent, requirements verification and auto-filling. We provide a systematic process for the creation of customizable and reusable templates, which, to the best of our knowledge, represents a novel contribution. The application of this systematic process to a subset of the ARINC-653 standard from the avionics domain, resulted in seven templates that cover various types of requirements. We developed a tool, called MD-RSuT, that supports the specification of requirements using the seven templates created for ARINC-653, and the automated verification and auto-filling of requirements using an ARINC domain model. We evaluated the applicability of the approach across domains, and its effectiveness in improving requirements quality in terms of necessity, unambiguity, completeness, singularity, and verifiability. To do so, we applied the approach on three case studies coming from different domains, namely avionics, automotive, and general purpose software. This evaluation encompasses over a thousand requirements. We also performed a quantitative and qualitative analysis of the results. The results show that RESPECT is applicable across domains, and it yields requirements with higher quality. [ABSTRACT FROM AUTHOR]
Datenbank: Academic Search Index
Beschreibung
Abstract:Requirements specification and verification are crucial processes of software development. These processes are particularly costly for safety critical systems due to the high number of requirements and their complexity. For such systems, it is important to use natural language for the specification, as requirements need to be readable by non-technical stakeholders and certification agents. To mitigate the inherent ambiguity caused by the use of natural language, controlled natural languages (CNL) are introduced as a means to constrain the specification while maintaining readability. In this paper, we leverage model-driven engineering (MDE) to propose RESPECT, REquirements SPECification using Templates, a CNL-based approach for requirements specification and verification. The fundamental idea of RESPECT is to use MDE techniques to: 1) model requirements' templates and thus ease their creation, implementation and evolution and 2) link the template models to existing domain models to support, to some extent, requirements verification and auto-filling. We provide a systematic process for the creation of customizable and reusable templates, which, to the best of our knowledge, represents a novel contribution. The application of this systematic process to a subset of the ARINC-653 standard from the avionics domain, resulted in seven templates that cover various types of requirements. We developed a tool, called MD-RSuT, that supports the specification of requirements using the seven templates created for ARINC-653, and the automated verification and auto-filling of requirements using an ARINC domain model. We evaluated the applicability of the approach across domains, and its effectiveness in improving requirements quality in terms of necessity, unambiguity, completeness, singularity, and verifiability. To do so, we applied the approach on three case studies coming from different domains, namely avionics, automotive, and general purpose software. This evaluation encompasses over a thousand requirements. We also performed a quantitative and qualitative analysis of the results. The results show that RESPECT is applicable across domains, and it yields requirements with higher quality. [ABSTRACT FROM AUTHOR]
ISSN:16191366
DOI:10.1007/s10270-025-01265-6