Research on Software Requirements Elicitation and Evolution using Repository Technology

1)从具有版本控制功能的Meta模型仓库生成的技术。我们开发了一种从软件图的Meta模型描述（如UML）生成仓库和CASE工具的技术。生成的存储库和工具可以具有版本控制机制，其中Meta模型的逻辑单元是变更管理的原子元素。相邻版本之间的差异（称为增量）被提取并存储到存储库中。此外，我们开发了一种技术的配置管理保持一致性时，集成和修改不同类型的模型片段的一部分的需求规范。该技术通过对Meta模型的约束来保持和检测增量操作记录中的不一致性; 2）使用本体为需求提供语义的技术和需求获取方法作为其应用，需要对规格说明文档进行语义分析以分析记录 ...更多信息 的变化要求。在这种方法中，我们采用了轻量级的语义处理技术，使用本体。将需求描述映射为本体元素，并在本体系统上进行推理以推断描述的属性。在此基础上，提出了一种需求获取方法及其支持工具，为用户提供了对缺失需求和不一致部分的建议。3）对规格说明的差异进行提取和分析，并将其抽象为模式的技术。我们开发了一个支持工具来提取映射到本体的需求变更的增量，并在语义上将它们分类。属于某个语义范畴的三角洲的大频率使我们能够组成进化模式，以预测未来的变化。例如，通过收集真实的变更记录，我们发现了一种模式，即功能性需求被修改，随后非功能性需求被修改。此外，我们开发了一种技术来分类的三角洲导致不一致或缺失的需求，并构建了一个自动化的系统，建议他们的候选人的需求文档。少

1)Technique to generate from meta models repositories having version control functions.We developed a technique to generate repositories and CASE tools from meta model descriptions of software diagrams like UML. The generated repository and the tool can have version control mechanism where logical units of the meta model are atomic elements of change management. The differences (called deltas) between adjacent versions are extracted and stored into the repository. In addition, we developed a technique of configuration management for keeping consistency when integrating and modifying different types of model fragments as parts of a requirements specification. In this technique, we specify the constraints on meta models for keeping and detecting inconsistency from operational records of deltas.2)Technique to provide semantics for requirements using ontology and a requirements elicitation method as its application.Semantic analysis of specification documents is necessary to analyze record … More s of changes of requirements. In this approach, we employed a technique of lightweight semantics processing using ontology. The descriptions of requirements are mapped into ontological elements and inferences to deduce the properties of the descriptions are possible on the ontological system. We also developed a requirements elicitation method and its supporting tool where guidance to suggest missing requirements and inconsistent parts is provided to users.3)Techniques to extract and analyze differences of specifications, and to abstract them into patterns.Based on the research results of the above 1) and 2), we developed a supporting tool to extract the deltas of requirements changes that are mapped to the ontology and to semantically classify them into categories. The large frequency of the deltas belonging to a certain semantic category allowed us to compose evolution patterns so as to anticipate future changes. By collecting real change records, for example, we found a pattern where functional requirements were modified and subsequently their non-functional ones were modified. Furthermore, we developed a technique to classify the deltas leading to inconsistent or to missing requirements and constructed an automated system to suggest their candidates out of a requirements document. Less