The Flow: From Deep-Learning to Digital Analysis and their Role in the HumanitiesCreating, Evaluating, and Critiquing Workflows for Historical Corpora

目前，只有具有深入的信息科学知识和编码能力的学者才能使用明确的工作流程和现成的工具来实施自然语言处理衍生的文本识别办法和方法。当专注于前现代文字和非西方语言时，这个问题甚至变得更加突出，因为在这些语言和非西方语言中，学者们需要对语言和所用技术的知识进行有效的理解。该项目将通过开发建立在现有技术基础上的数字工作流程来解决这一问题，从而产生处理历史资料的新方法。在过去的十年里，机器学习方法已经出现，允许以全新的方式处理文本。然而，由于这些技术的不透明性和复杂性，我们只能看到它们对人文学科的影响。通过我们的项目，我们将进一步在人文学科中更广泛地采用和关键使用机器学习技术，采用它们来运作理论观点，并在考虑到历史问题和伊斯兰研究的情况下对技术进行评估，重点放在法律和行政语料库上。13-15世纪的英格兰，14-17世纪的北欧汉斯地区，16-18世纪的瑞士和19世纪的埃塞俄比亚的法律和行政来源将在行为学和制度框架内进行分析。这一领域的研究问题得益于数字工作流程的发展，特别是因为它们侧重于分析社会进程、法律如何实践以及它如何影响日常生活的方方面面。这些过程通常隐含在相关文档中，这使得它们很难被发现。本项目中的工作流解决了这个问题，并有助于使看不见的东西变得可见。因此，我们将能够分析每个时期和地区各自的法律实践，完善我们对稳定以及法律实践变化的理解，甚至能够对社区的社会生活产生洞察，否则很难描述这些社会生活。

At the moment, clear-cut workflows and ready-made tools for the implementation of text recognition approaches and methods derived from natural language processing are only available to scholars with in-depth knowledge of information sciences and the ability to code. This problem gets even accentuated, when focusing on pre-modern scripts and non-western languages, where scholars need a working understanding of the language and knowledge about the technology used. This project will address this issue by developing digital workflows that build on available technologies, leading to new means of working with historical sources. In the past decade, machine-learning approaches have been made available that allow for the processing of text in completely new ways. Due to the opacity and complexity of these technologies, however, we see their impact only at the margins of the humanities. With our project, we will further the wider adoption and critical use of machine-learning technologies in the humanities, adopting them to operationalize theoretical perspectives and evaluating the technologies with questions of history and Islamic studies in mind focusing on legal and administrative corpora. Legal and administrative sources from the 13-15th centuries in England, the 14th-17th centuries in the northern European Hanse area, from the 16-18th centuries in Switzerland and the 19th century in Ethiopia will be analyzed within a praxeological and institutional framework. Research questions in this area profit from the development of digital workflows in particular because of their focus on the analysis of societal processes, of how law has been practiced and how it affected aspects of everyday life. These processes are usually implicitly present in the relevant documents and this makes them hard to detect. Workflows such as the one in this project address this problem and help to make the invisible visible. We will therefore be able to analyze the respective legal practices of each period and place, refine our understanding of the stabilization as well as the changes in legal practices, and even generate insights in the social life of communities that would otherwise be hard to depict.