Requirements definition methane-sad on comparison among existing systems and kncwkdge ofStakeholders

信息系统嵌入在组织的特定业务或活动中，因此，关于此类业务和组织的知识，即领域知识，在此类系统的需求分析中至关重要。因此，在需求分析中，有效、高效地利用领域知识是非常重要的。为了有效、高效地利用领域知识文档，不仅内容重要，表示方式也很重要。此外，使用这些领域知识的利益相关者的知识也很重要。在本研究中，我们报告了领域知识的创造方法、提供这些知识的方法、需求分析过程中不同表示的影响以及需求定义过程中领域知识的个性化。我们开发并评估了一种基于现有类似系统文档的领域知识创建方法。基于领域知识，本文提出了一种类似新系统的需求定义方法。为了研究如何有效地显示这些领域知识，我们设计了四个度量需求规格说明效果的度量标准。通过使用度量标准，我们做了比较实验来评估对需求文档质量表示的影响。我们分析了两个需求文档，每个需求文档都是通过使用不同的领域知识表示来构建的：一个是字典样式，另一个是UML样式。因此，字典样式似乎适合于全面的描述，而UML样式似乎适合于对领域知识的原样重用。我们定义了一个识别领域知识难度的度量和一个利益相关者专业知识的度量。在实验研究中，我们证实了这两个度量有助于领域知识的个性化。最后，我们基于上面的结果开发了一个支持需求定义的原型工具。少

An information system is embedded in a specific business or activities in an organization, thus, the knowledge about such a business and an organization, i.e. domain knowledge, is crucially important in a requirements analysis for such a system. Therefore, the effective and efficient usage of domain knowledge is very important in requirements analysis.In order to use domain knowledge documents effectively and efficiently, not only the contents but also the representation is important. In addition, knowledge of stakeholder who uses such domain knowledge is also important. In this research, we report the results about the method for domain knowledge creation, the method for providing such knowledge, the effect of different representations during requirements analysis, and personalization for domain knowledge during requirements definition.We developed and evaluated a method for domain knowledge creation based on documents about existing similar systems.Based on the domain knowledge, we p … More roposed a requirements definition method for a similar new system.To investigate how to show such domain knowledge effectively, we designed four metrics that measure the effect of the requirements specification. By using the metrics, we did comparative experiments that evaluate the effect of the representations on the quality of requirements documents. We analyzed two requirements documents each of which is constructed by using different domain knowledge representations: one is dictionary style and the other is UML style. As a result, dictionary style seemed to be adequate for the comprehensive description and UML style seemed to be adequate for the as-is reuse of the domain knowledge.We defined a measure for identifying difficulty of domain knowledge and another measure for expertise of a stakeholder. During a experimental study, we confirmed these two measures contribute to personalization for domain knowledge. Finally, we developed a prototype tool for supporting requirements definition based on the results above. Less