Semantically configurable software modelling notations and tools

操作软件中的错误对用户和软件开发人员都产生了重大影响，需要进行返工以消除错误的影响，需要进行额外的工作以避免或减轻错误，需要由于服务关闭或延迟而失去业务，需要增加客户支持。这些错误大多是由于软件要做什么的初始规范中的遗漏、模糊、不一致或错误，而不是在开发过程中引入的设计或编码错误。此外，规范错误发现得越晚，纠正它们的代价就越高。我的大部分研究都集中在建模和分析软件规范上，以便在软件开发的早期阶段检测错误，此时错误更便宜，也更容易修复。在过去的几年里，我的研究小组一直致力于开发支持多种建模符号的语义可配置工具。我们已经开发了一种构造建模符号定义的方法，这样定义就可以被解释并用于配置建模工具，例如模型编辑器和模型验证器。继续这项研究，我计划研究如何简化配置新建模符号定义的任务。我还打算调查我们的工作是否可以适应异构模型，从而使说明者能够组合用不同建模符号编写的模型。此外，我计划探索优化语义可配置工具的性能的方法。这项研究将使软件说明者能够创建他们自己的建模符号——与他们正在建模的问题相匹配，领域专家可以阅读和验证——但仍然可以使用传统工具来编辑、操纵和检查他们的模型。这种技术最终将导致软件的改进。

Errors in operational software significantly impact both users and software developers, in terms of rework to undo the effects of errors, extra work to avoid or mitigate errors, lost business due to shutdowns or delays in services, and increased customer support. Most of these errors are due to omissions, ambiguities, inconsistencies, or mistakes in the initial specification of what the software is to do, as opposed to design or coding errors introduced during development. Moreover, specification errors become more expensive to correct the later they are detected.Much of my research focuses on modelling and analyzing software specifications to detect errors in the early phases of software development, when errors are cheaper and easier to fix. For the past few years, my research group has been working on developing semantically configurable tools that support multiple modelling notations.  We have developed a way of structuring the definition of a modelling notation such that the definition can be interpreted and used to configure modelling tools, such as model editors and model verifiers.Continuing this research, I plan to study how to ease the task of configuring the definition of a new modelling notation.  I also intend to investigate whether our work can be adapted to accommodate heterogeneous models, thereby enabling specifiers to combine models written in different modelling notations.  In addition, I plan to explore ways of optimizing the performance of semantically configurable tools.This research will enable software specifiers to create their own modelling notations - that match the problems they are modelling and that domain experts can read and validate - but still have access to traditional tools for editing, manipulating, and checking their models.  Such technology will ultimately lead to improved software.