Unpacking 'in-the-Wild' Design and Engineering Practices

对话用户界面(CUI)早在1960年就以基于文本的对话系统的形式出现，但人们对它的兴趣越来越大，特别是在2014年语音驱动的“助手”亚马逊Echo发布之后。人机界面和CUI的具体研究已经开始强调语音助手和其他平台设计的作用，越来越多的人对建立CUI的设计框架、模式、指导、启发式或原则感兴趣。与此同时，实践者在借鉴自己在新平台和新技术方面的经验的同时，正在迅速发展自己的设计知识体系。但实践和研究之间的距离是很大的。研究促进了复杂的设计框架和模式，这些框架和模式旨在为设计和工程CUI提供可重用的方法，但从业者变幻莫测的工作与现有平台(例如，AWS Lex、IBM Watson Assistant等)紧密结合(这是可以理解的)。此外，新的对话式设计实践和工程的情况目前很少有文献记载。了解这些条件对于基础研究人员在日常实践中设计和实施CUI的设计和工程框架至关重要。本博士通过与参与设计和工程基于语音和聊天的CUI的人员进行人种学实地工作来正面解决这个问题，以揭示和阐明从业者(及其利益相关者)设计和实施CUI以进行生产级部署的(协作)方式。通过与CUI研究社区分享这些知识以及让从业者参与研究成果，作为本论文工作的成果，博士将在社区之间建立双向参与。

Conversational User Interfaces (CUIs) have been around since as early as 1960 in the form of text-based dialogue systems but there is significant new interest in them, particularly since release of the voice-driven 'assistant' Amazon Echo in 2014. HCI and CUI specific research has begun to emphasise the role of design for voice assistants and other platforms, increasing interest in establishing design frameworks, patterns, guidance, heuristics or principles for CUIs. At the same time, practitioners - while drawing upon their own experiences with new platforms and technologies - are rapidly developing their own body of design knowledge.But the distance between practice and research is significant. Research promotes sophisticated design frameworks and patterns which aim to offer reusable methods for design and engineering CUIs, yet the vagaries of practitioners' work is strongly wedded (understandably) to existing platforms (e.g., AWS Lex, IBM Watson Assistant, etc.). Further, the conditions of new conversational design practice and engineering are currently little documented. Understanding those conditions is vital to ground researchers' design and engineering frameworks for CUIs in the realities of everyday practice.This PhD tackles this problem head-on by engaging in ethnographic fieldwork with those involved in designing and engineering voice and chat based CUIs, to uncover and articulate the (collaborative) ways in which practitioners (and their stakeholders) design and implement CUIs for production-level deployment. By sharing this knowledge with CUI research communities as well as engaging practitioners in the research outputs, the PhD will build engagement 'both ways' between communities as a product of this thesis work.