MDE-Net: A UK Expert Network in Model-Driven Engineering

我们面临着一场危机：在未来数年，我们需要开发的软件，将远远超过我们所能培养的训练有素的软件工程师所能处理的数量。到2020年，整个欧洲将缺少525，000至750，000名合格的IT专业人员。模型驱动软件工程（MDE）使具有领域知识但没有软件专业知识的专家能够利用软件工程师的工作，并成为他们使用的软件的创建的积极贡献者。在MDE中，软件不是用通用编程语言（GPL）（如C++）开发的，而是使用定制的、特定于领域的语言（DSL），这些语言是根据领域专家的需求开发的。这样的DSL使用领域专家熟悉的术语和语法，但由编译，分析和解释基础设施支持，允许DSL中的语句由计算机自动评估，类似于用GPL编写的软件。这就实现了关注点的分离：领域专家可以贡献他们的专业知识并做出与领域相关的决策（例如，关于计算设备购买税的具体规则应该是什么），而不必考虑有效计算的技术含义。与此同时，软件开发人员可以专注于为特定领域的效率和可扩展性提供支持，而不必担心特定领域问题的详细规范。英国在MDE研究方面具有相当的实力。然而，渗透到实践中仍然缓慢，MDE研究不容易被个别实验室以外的人所获得。造成这种情况的原因是，MDE提供的机会缺乏可见性，缺乏易于获取和查找的培训材料，以及缺乏围绕MDE的组织良好的社区。这也使得它难以识别和追求新的研究方向在该领域，特别是在学术界和工业界之间的合作。MDE-Net将是第一个英国网络模型驱动的软件工程。EPSRC网络是应对上述挑战的最佳途径：该网络将提供有关MDE的中央信息源，包括成功案例、培训材料、定期更新的技术雷达，突出显示研究和实践的现状，以及英国MDE专家目录。该网络将成为MDE社区的焦点，使学术研究人员和工业实践者之间能够进行密切和持续的互动，并推动交叉融合到其他通常不会从软件工程角度考虑的学科（例如，MDE-Net将由伦敦国王学院的一个团队领导，由Steffen Zschaler博士（国王学院）和Perdita Stevens教授（爱丁堡）领导。将通过由网络成员组成的3个工作组开展工作。WG 1将通过定期举办研讨会，将MDE的研究人员与行业和其他学科的相关利益相关者聚集在一起，致力于推动MDE未来有影响力的研究。第一工作组还将管理一个基金，以支持种子玉米项目，以便建立新的研究合作。第二工作组将侧重于培训和教育，包括整理现有的和委托编写新的培训材料，以及提供定期的培训活动方案和促进课程开发。最后，WG 3将致力于围绕MDE建立一个更加一致的社区，并提高MDE成功和机会的可见度。除了通过网络网站提供信息外，还将通过制作和定期更新MDE技术雷达以及举办年度MDE国家研讨会来实现这一目标。该网络的目标是在供资期结束前实现自给自足。

We are facing a crisis: over the coming years, substantially more software needs to be developed than can be handled by the number of trained software engineers we can produce. Across Europe, we lack between 525,000 and 750,000 qualified IT professionals by 2020. Model-driven software engineering (MDE) enables experts with domain knowledge but without software expertise to leverage the work of software engineers and become active contributors to the creation of the software they use. In MDE, software is developed not in general-purpose programming languages (GPL) like C++, but using bespoke, domain-specific languages (DSLs) that are developed with domain experts' needs in mind. Such DSLs use terminology and syntax that is familiar to domain experts, but are underpinned by compilation, analysis, and interpretation infrastructure that allows statements in a DSL to be automatically evaluated by a computer, similar to software written in a GPLs. This enables separation of concerns: domain experts can contribute their expertise and make domain-relevant decisions (for example, about what the specific rule should be for computing tax on equipment purchases) without having to consider the technical implications of efficient computation. At the same time, software developers can focus on providing support for efficiency and scalability for a particular domain without having to worry about the detailed specification of specific domain problems.The UK has a substantial strength in MDE research. However, penetration into practice is still slow and MDE research is not easily accessible to those outside of individual labs. The reasons for this situation can be found in a lack of visibility of opportunities enabled by MDE, a lack of easily accessible and easy to find training materials, and a lack of a well-organised community around MDE. This also makes it difficult to identify and pursue new research directions in the field, especially in collaboration between academia and industry.MDE-Net will be the first UK network on model-driven software engineering. An EPSRC network is the best way to address the challenges described above: the network will provide a central source of information about MDE, including success stories, training materials, a regularly updated technology radar highlighting the current state of research and practice, and a catalogue of UK experts in MDE. The network will become the focal point of the MDE community, enabling close and continuous interaction between academic researchers and industrial practitioners as well as driving cross-fertilisation into other disciplines that are not typically considered from a software engineering perspective (e.g., biology, artificial intelligence, robotics, data science, urban sciences, geography, or medicine).MDE-Net will be led by a team hosted at King's College London and led by Dr Steffen Zschaler (King's) and Prof Perdita Stevens (Edinburgh). Work will be undertaken through 3 working groups constituted from the network membership. WG1 will focus on driving impactful future research in MDE by bringing together researchers in MDE with relevant stakeholders from industry and other disciplines through a regular programme of workshops. WG1 will also manage a fund to support seed-corn projects so that new research collaborations can be built. WG2 will focus on training and education by curating existing and commissioning new training materials as well as offering a regular programme of training events and contributing to curriculum development. Finally, WG3 will work towards building a more coherent community around MDE and increasing the visibility of MDE successes and opportunities. Beyond making information available through the network website, this will be achieved by producing and regularly updating an MDE technology radar and running an annual national symposium on MDE. The network aims to achieve self-sufficiency by the end of the funding period.