Fatima: a digital platform to transform qualitative reasearch in marginalised sectors

定性研究将为有效的商业活动、非政府组织干预和政府政策提供基础，以应对COVID-19的直接影响，并更广泛地为英国更广泛的社会经济复苏提供支持。然而，尽管老年人、社会经济弱势群体和融合程度较低的族裔人口受到的影响最大，但COVID-19进一步加剧了这些边缘化且往往脆弱的群体本已面临的定性研究挑战。由于社交距离和旅行限制阻碍了最有效的面对面数据收集（包括访谈、焦点小组和实地工作），在这些群体中远程开展定性研究面临着可及性和规模方面的根本障碍，原因是：*无法获得与大多数连接平台兼容的宽带、计算机和智能手机；数字素养总体较低*缺乏管理受访者数据收集的技术，这些受访者只能通过固定电话或“非智能”移动功能手机联系，并且跨多个平台整合有效的研究工作流程*研究设计和灵活性，以提取复杂和敏感的多维情绪数据；考虑到这些受访者的具体情况——这些边缘化和弱势群体的研究人员的许多目标也可能由于保护或其他原因而削弱了表达直接反应的能力*昂贵的人工密集型分析方法的可扩展性，以便在更复杂的主题模型和其他隐藏的定性数据变量中清楚地识别和关联这些依赖关系尽管对可操作输出的需求增加，因此，这项工作在很大程度上被商业活跃和垂直整合的研究机构暂停了。现有的用于定性数据收集的数字工具集是以消费者为中心的基于网络/智能手机的应用程序，主要是为了满足大型企业市场研究人员的需求而设计的。这些通常使用自动语音到文本的转录，但只支持非常低层次的情感分析，用于简单的严格约束的研究场景。作为回应，Maido正在开发Fatima，这是一个定性数据收集和分析的数字平台，它独特地连接了端到端的研究价值链，为那些只能通过功能手机对话访问的边缘化受访者提供服务。Maido已经完成了一个成功的概念验证开发，实现了记录数据收集的核心数据架构和远程工作流程。利用他们的最小可行产品和经过验证的转录和自然语言处理工具集，Maido现在寻求将定性特征和编程功能完全开发成一个可扩展的全平台原型，具有高水平的分析自动化。

Qualitative research will underpin effective commercial activities, NGO interventions and government policies to address the direct impacts of COVID-19, and more generally for the UK's wider socio-economic recovery.However, whilst the elderly, socio-economically disadvantaged and less-integrated ethnic demographics are impacted most, COVID-19 has further increased the already unmet challenge of qualitative research within these marginalised and often vulnerable groups.With the most effective face-to-face data gathering - including interviews, focus groups and field work - prevented by social distancing and travel restrictions, remotely undertaking qualitative research within these groups faces fundamental barriers to accessibility and scale due to:* Poor access to broadband, computers and smartphones compatible with most connectivity platforms, and overall low rates of digital literacy* Lack of technology to manage data gathering from respondents who can only be reached via landline or 'non-Smart' mobile feature phones, and integrating effective research workflows across multiple platforms* Study design and agility to extract the complex and sensitive multi-dimensional sentiment data, given these respondents' specific scenarios -- many targets for researchers in these marginalised and vulnerable groups may also have compromised capacity to articulate direct responses for reasons of safeguarding or otherwise* Scalability of costly manually-intensive analysis methodologies to clearly identify and correlate these dependencies across more complex topic models and other hidden qualitative data variablesDespite an increased need for actionable outputs, this work has therefore largely been suspended, both by commercially-active and vertically-integrated research organisations.Existing digital toolsets for qualitative data gathering are consumer-focussed web/smartphone-based applications, which are predominantly designed to meet the needs of large-scale corporate market researchers. These commonly use automated speech-to-text transcription, but support only very low-level sentiment analysis for simple tightly-constrained research scenarios.In response, Maido are developing Fatima, a digital platform for qualitative data collection and analysis, which uniquely connects the end-to-end research value chain for marginalised respondents who can only be accessed via feature phone conversations.Maido have already completed a successful proof-of-concept development, realising the core data architecture and remote workflows for recorded data gathering. Leveraging their minimum-viable-product and proven transcription and natural language processing toolsets, Maido now seek to fully develop the qualitative features and programmatic functionality into a scaled full-platform prototype, with a high level of analytical automation.