A method for creating semantic thesauri of requirements using the SKOS ontology

Abstract Problems related to requirements specification, such as ambiguity and incompleteness, are still recurrent within information system development processes. Reusing requirements is one of the mechanisms that can help reduce these setbacks. In this sense, this paper aims to propose a method for creating and publishing reuse-oriented semantic thesauri of requirements, using the Semantic Web’s technologies and standards, and following the principles of Linked Data. The central ontology used for a formal description of these thesauri is the Simple Knowledge Organization System. This ontological model provides a set of thesaurus-creating axioms and properties, allowing accurate and reliable documentation of the definition, hierarchy, and other interrelationships among a system’s requirements, in a knowledge graph. We present a prototypical Web service that functions as a repository for reuse and demonstrates the method in practice. We also describe a study about the feasibility of the proposal, which was carried out with professionals and included a group discussion and later, the completion of individual evaluation questionnaires. The study’s results were mostly favorable, and some suggestions for improvement were appointed. The participants considered the proposal relevant to Requirements Engineering and with potential for expansion, since the guidelines presented allow the creation of new kinds of inference and navigability over the stored requirements.

摘要：在信息系统开发流程中，需求规格说明（requirements specification）相关的歧义性、不完整性等问题仍时有发生。复用需求是缓解此类弊端的有效途径之一。有鉴于此，本文旨在提出一种面向需求复用的语义叙词表（semantic thesauri）构建与发布方法，该方法采用语义网（Semantic Web）技术与标准，并遵循关联数据（Linked Data）的设计原则。用于对这类叙词表进行形式化描述的核心本体为简单知识组织系统（Simple Knowledge Organization System）。该本体模型提供了一系列叙词表构建公理与属性，能够在知识图谱（knowledge graph）中精准且可靠地记录系统需求的定义、层级结构以及其他相互关联关系。本文提出了一款作为复用知识库的原型Web服务，并通过该实例验证了所提方法的实际应用效果。此外，本文还介绍了一项针对该方案可行性的实证研究：研究对象为领域专业人士，研究环节包含小组讨论与后续的个人评估问卷填写工作。该研究的整体结果较为积极，同时也收集到了若干改进建议。参与人员认为，该方案与需求工程（Requirements Engineering）领域高度相关，且具备拓展潜力——因其所提出的指导原则可支持对存储的需求开展新型推理与导航操作。