Using Contextual Information to Reduce Changes in Information Systems Requirements

管理不断变化的用户需求是开发信息系统的一个持久而重大的挑战。在这项研究中，我们提出了一种预测这种需求的方法，而不是试图管理它们。为此，我们提出了一个在系统分析和设计中收集和利用相关上下文信息的框架。在本研究中，上下文是功能和非功能需求规范中不可用(或不需要)的任何信息，但对以下方面有用：确定需求的含义，检测之前未确定的整个需求集，以及解决需求的潜在歧义。我们的假设是：将用户和系统分析人员置于正确的上下文中(即使只是部分上下文)将生成更准确、更完整的需求，从而减少对用户需求的更改，这通常是代价高昂的。*为了开发框架，我们将首先调查和调整在相关科学领域中开发的上下文表示法，以用于系统分析和设计。这些包括人工智能(例如，知识表示和智能代理)、组织科学(例如，组织结构、组织行为和组织学习)、科学哲学(例如，提供上下文信息的叙述)和学习(例如，认知科学和心理学理论)。我们将采用设计科学的方法，因为我们的目标是：开发一个可行的人工制品(框架)；并在现实世界的问题上对其进行评估和测试。这将需要一种迭代的方法，采用相关科学领域的理论和研究成果。*拟议研究的总体思路类似于在人机交互、物联网、智能家居、机器人和无处不在计算中使用的上下文信息。在这些领域开展的工作将被用作投入。然而，在系统分析中使用上下文信息与其他领域并不完全相同，因为系统分析的上下文包括组织行为、组织学习和复杂的业务流程，这超出了个人在环境中工作的范围。在基于目标的需求工程、用例和用户故事中也开展了重要的相关工作，它们可以提供需求所需的一些上下文信息。这项研究将通过包括更多的、更以业务和组织为中心的上下文来增加这一知识库。*鉴于成功的信息系统项目数量很少(据https://www.infoq.com/articles/standish-chaos-2015)，报道，约有30%)，这项研究的潜在影响很大，可以直接帮助组织节省成本。

A persistent, and major, challenge in developing information systems is managing changing user requirements. In this research, we propose a way to anticipate such requirements instead of trying to manage them. To do so, we propose a framework for gathering and utilizing relevant contextual information in systems analysis and design. Context, in this research, is any information that is not available (or not needed) in functional and non-functional requirements specification, but useful to: determine the meaning of the requirements, detect an entire set of requirements not previously identified, and resolve potential ambiguities of requirements. Our hypothesis is that: situating users and systems analysts in the right context (even if only a partial context) will generate more accurate, complete requirements, resulting in less changes to user requirements, which are often costly.******To develop the framework, we will first survey and adapt contextual representation developed in related scientific fields for the purposes of systems analysis and design. These include artificial intelligence (e.g., knowledge representation and intelligent agents), organization science (e.g., organizational structure, organizational behavior, and organizational learning), philosophy of science (e.g., narratives to provide contextual information), and learning (e.g., cognitive science and psychology theories). A design science approach will be adopted because our objectives are to: develop a workable artifact (the framework); and evaluate and test it on real world problems. This will require an iterative approach, adopting theories and research results from the related scientific fields.******The general idea of the proposed research is similar to contextual information used in human-computer interactions, the Internet of things, smart homes, robotics, and ubiquitous computing. Work performed in these areas will be used as input. However, employing contextual information in systems analysis is not exactly the same as other areas because the context of systems analysis includes organizational behavior, organizational learning, and complex business processes, which go beyond individuals working in an environment. There has also been significant, relevant work carried out in goal-based requirements engineering, use cases, and user stories that can provide some of the contextual information needed for requirements. This research will add to this knowledge base by including additional, more business and organization centric, context.******Given the small number of successful information systems projects (approximately 30% as reported at https://www.infoq.com/articles/standish-chaos-2015), the potential impact of this research is high and could directly contribute to organizational savings.