Knowledge infrastructures are defined as robust networks of people, artifacts, and institutions that generate, share and maintain specific knowledge. Yet, many domains are fragmented and far from robustly networked, such as science communication or aerospace engineering. While FAIR (Findable, Accessible, Interoperable, Reusable) data management tools exist, their adoption in these domains is limited. Several challenges inhibit this adoption, from complex heterogeneous data formats to lack of structured support to outright incentives against collaboration or legal barriers. This doctoral work outlines how to foster underdeveloped knowledge infrastructures with the use-cases of science communication and aerospace engineering. By analyzing these problems and identifying available solutions, tool-supported workflows towards collaborative infrastructure can be implemented and evaluated. These include human-in-the-loop artificial intelligence (AI)-supported workflows for information extraction and processing, wiki- and knowledge-graph-based digital libraries, and stakeholder-requirement-driven interfaces. While these developed tools for workflow automation and knowledge representation show promise, significant challenges remain. Future work will have to go beyond technical problem-solving and address the societal and legal barriers to unlock the particular domains. Beyond that, advocates of emerging knowledge infrastructures in any domain are welcome to apply the findings of this work to foster the networking of available knowledge.

翻译：知识基础设施被定义为由人员、制品和机构组成的稳健网络，用于生成、共享和维护特定知识。然而，许多领域（如科学传播或航空航天工程）仍处于碎片化状态，远未形成稳健的网络化体系。尽管存在FAIR（可发现、可访问、可互操作、可重用）数据管理工具，但它们在上述领域的应用十分有限。从复杂的异构数据格式、缺乏结构化支持，到合作动机的缺失乃至法律障碍，多重因素阻碍了这些技术的采纳。本博士论文以科学传播和航空航天工程为应用场景，探讨如何培育欠发达的知识基础设施。通过分析这些问题并识别现有解决方案，可以实施并评估面向协作基础设施的工具支持型工作流。这些工作流包括：基于人在回路的人工智能（AI）支持的信息抽取与处理流程、基于wiki和知识图谱的数字图书馆，以及面向利益相关者需求的交互界面。尽管开发的这些工作流自动化与知识表示工具展现出潜力，但重大挑战依然存在。未来的工作需超越技术层面的问题解决，致力于消除特定领域的社会与法律障碍。此外，任何领域新兴知识基础设施的倡导者均可借鉴本研究成果，以促进现有知识的网络化整合。