Unlocking our Digital Past with Artificial Intelligence (LUSTRE)

越来越多的政府数据以数字形式创建。电子邮件取代了信件，PDF和Word文档取代了纸质备忘录，音频/视频文件存储在政府内部档案和各种系统中。然而，这些数据中只有一小部分被转移到国家档案馆和其他档案库，用于长期保存、访问和使用。Lustre项目旨在通过将政府专业人员与计算机科学家、数字人文主义者和文化遗产组织的档案保管员联系起来，解锁这些数据。它将侧重于将人工智能(AI)应用于数字档案记录，以使其更容易获取。事实上，人工智能可以用于敏感性审查(即在海量数据中识别敏感文档)，从而有可能发布非机密的记录。人工智能还可以用来搜索海量数据。但至关重要的是，在选择和处理数据时要避免偏见，这可能会歧视某些群体，甚至影响集体记忆。这就要求政策制定者与算法打交道，而不是将人工智能视为“黑匣子”。在数字革命之后，无法获取政府记录的问题变得尤为严重。用来组织纸质记录的严格的归档系统。然而，这些系统并不能很好地适应数字时代。2017年，由内阁办公室和国家档案馆联合撰写的《为更好的政府提供更好的信息》报告指出了政府内部出生数字记录管理方面的问题--包括组织不善的记录、分散在不同系统中的记录以及几乎不可能有效的搜索。这种缺乏组织的情况导致在查找信息和访问用户需要的记录方面存在困难。与生俱来的数字记录的规模也使搜索信息变得极其复杂，特别是当数据分散在多个设备和系统上时。这些数据可能包含机密和敏感材料，包括可能对恐怖分子和其他对手有用的材料。为了限制风险，数据通常被锁定，用户--包括历史学家、社会科学家、记者和第三部门专业人员--无法访问。档案是用来使用的，而不是锁起来的。无法获取的政府记录在短期内导致缺乏问责，并有可能在长期内影响文化记忆。我们如何改进以数字形式获取政府档案记录的途径？LUSTRE项目旨在通过提供以下产出来解锁这些数据：--在内阁办公室举行4次午餐会谈；--共举办4次面对面讲习班，包括在伦敦举办的3个讲习班(内阁办公室和科学博物馆)和在贝尔法斯特举办的一个讲习班(由北爱尔兰历史记录办公室主办)；--在线调查和50次半结构化访谈；--开放获取报告和期刊特刊，包括一篇由主计委和博士后共同撰写的文章；_关于天生数字档案的跨部门网络，将政府专业人员与学者和魅力专业人士联系起来。网站、相关的社交媒体和专用的LUSTER LIST-SERV将帮助我们与政府、学术界、档案机构和其他方面的相关方联系起来。

More and more government data are created in digital form. Emails have replaced letters, PDFs and Word documents have replaced paper memos, and audio/visual files are stored in governmental internal archives and in various systems. Yet just a small proportion of these data is transferred to The National Archives and other archival repositories for long-term preservation, access and use. The LUSTRE project aims to unlock these data by connecting government professionals with Computer Scientists, Digital Humanists and archivists in cultural heritage organisations. It will focus on the application of Artificial Intelligence (AI) to digital archival records in order to make them more accessible. Indeed, AI can be used for sensitivity review (i.e., to identify sensitive documents in a mass of data), making it possible to release records that are not confidential. AI can also be used to search vast amounts of data. But it is crucial to avoid biases in the selection and processing of data, which could discriminate against certain groups and even impact the collective memory. This requires policy makers to engage with algorithms rather than treating AI as a "black box."The problem of inaccessible governmental records has become particularly acute following the digital revolution. Rigorous filing systems used to organise paper records. However, these systems are not well adapted to the digital age. In 2017, the report Better Information for Better Government (co-authored by the Cabinet Office and The National Archives) identified issues with the management of born-digital records within government - including poorly organised records, scattered across different systems and almost impossible to search effectively. This lack of organisation leads to difficulties in finding information and giving access to records that users need. The scale of born-digital records also makes it extremely complicated to search for information, particularly when data are scattered on multiple devices and systems. These data could contain confidential and sensitive materials, including materials that could potentially be useful to terrorists and other adversaries. In order to limit risk, data is often locked away and inaccessible to users - including historians, social scientists, journalists and third sector professionals. Archives are meant to be used, not locked away. Inaccessible government records lead to a lack of accountability in the short term, and risk impacting the cultural memory in the long term. How can we improve access to government archival records in digital form? The LUSTRE project aims to unlock these data by delivering the following outputs:_4 lunchtime talks at the Cabinet Office; _a total of 4 face-to-face workshops, including three workshops in London (Cabinet Office and Science Museum) and one workshop in Belfast (hosted by Public Records Office of Northern Ireland);_online survey and 50 semi-structured interviews; _open-access report and journal special issue, including one article co-authored by the PI and postdoc;_cross-sector network on born-digital archives, connecting government professionals with academics and GLAM professionals.A website, associated social media, and a dedicated LUSTRE list-serv will help us connect with interested parties - in government, academia, archival institutions and beyond.