Software Innovation for Mass Emergencies

“给顾客他们想要的，顾客不知道他们想要什么。”然而，软件需求引出的主要来源是用户。假设涉众知道软件应该为他们做什么，研究团体通过分析与软件产品相关或关于软件产品的数据源，开发了有效的需求提取方法。这些资源要么是面试或头脑风暴会议的结果，要么是软件文档，要么是用户反馈。受需求工程的数据驱动愿景的启发，并相信用户不能精确地想象软件如何以及何时可以帮助他们，我们将需求引出的来源从关于软件的文档转移到关于主题、事件或问题的通用（与软件无关）通信。这些非结构化的通信可以是文本的（例如推文，公开论坛中的讨论）或非文本的（例如Instagram上分享的照片，飓风视频）。这意味着，如果一群人正在就一个事件进行交流，那么本提案中开发的方法可以创新，并提供帮助他们面对事件的软件服务。这可以是任何群体或社区，例如，大规模紧急情况中的受害者或志愿者，患有慢性疾病（如糖尿病）的人，或特定群体（如老年人或学生）。考虑到一个软件不适合所有关注给定主题的人，解决方案不能是一个产品。相反，我们的方法引入了一系列软件产品。我们调整和评估我们针对大规模突发事件提出的方法。移动设备在紧急情况下发挥着更大的作用，因为它们可能突然成为其所有者唯一的资源之一。对于软件工程师来说，这是一个为移动设备配备软件解决方案（也称为移动应用程序）的独特机会。在我读博士期间，我只关注麦克默里堡（Fort McMurray）野火期间Twitter上的文本交流，我发现，在现有的任何紧急应用程序中，前40个基本功能中有85%是不可用的。国内和国际媒体报道的研究结果引发了许多合作请求。在本提案中，我们遵循三个主要目标：-开发使用机器学习从文本和非文本通用通信中推断需求的方法。-开发一种自动化的方法来定义应用程序功能，以满足使用数据统计和挖掘方法的需求。-为大规模突发事件提供移动应用程序产品线，通过使用基于超启发式和元启发式搜索的技术优化了一组功能。这个跨学科的研究方案对人们的日常生活有切实的影响。我计划用这个关于加拿大社会相关问题的研究计划资助2名博士、2名硕士和5名本科生或高中生。

"Give the customers what they want, customers don't know what they want."(S. Jobs) Yet, the primary source for software requirement elicitation is users. Assuming that stakeholders know what the software should do for them, the research community developed efficient methods for requirement elicitation through analyzing data sources related to or about software products. These sources are either output of interview or brainstorming sessions, software documents, or user feedback. Inspired by the data-driven vision for requirements engineering and by believing that users cannot precisely imagine how and when software can help them, we shift the source of requirements elicitation from documents about software to general-purpose (and unrelated to software) communication about a topic, event, or a problem. These unstructured communications could be textual (e.g. tweets, discussions in open forums) or non-textual (e.g. photos shared on Instagram, videos of a hurricane). This means, if a group of people are communicating about an event, the methods developed in this proposal can innovate and offer a software service helping them facing the event. This could be any group or community, for example, victims or volunteers in a mass emergency, people suffering a chronic disease (such as diabetes), or specific groups (like senior citizens, or school students). Considering that one software does not fit all the people being concerned with a given topic, the solution cannot be one product. Instead, our methods introduce a family (line) of software products. We tune and evaluate our proposed methods toward mass emergencies. A mobile device takes on a heightened role in emergency situations, as they may suddenly be among their owner's only resources. This is a unique opportunity for software engineers to equip mobile devices with a software solution (also known as a mobile app). During my PhD and only by focusing on textual communication in Twitter during the Fort McMurray wildfire, I showed that 85% of the top 40 essential features were not available in any of the existing emergency apps. The research results covered in national and international media triggered numerous collaboration requests. In this proposal we follow three main objectives: - Developing methods to infer requirements from textual and non-textual general-purpose communications using machine learning. - Developing an automated method for defining app features for satisfying requirements using data statistics and mining methods. - Offering a mobile app product line for mass emergencies with an optimized set of features from using hyper and metaheuristic search-based techniques. This cross-disciplinary research proposal has tangible impact on the daily life of people. I plan to fund 2 PhD, 2 Masters, and 5 undergraduate or high-school students with this research proposal on problems relevant to Canadian society.