Speaker
Description
This poster delves into the critical role of authority files in advancing AI-based knowledge management, focusing on the context of the German National Library's (DNB) efforts. Authority files of the Gemeinsame Normdatei (https://gnd.network/Webs/gnd/EN/Home/home_node.html) serve as the cornerstone for successful information retrieval, interoperability, and the overall effectiveness of AI applications, such as chatbots. At the DNB for instance we explore how leveraging curated and high-quality datasets from library collections can enhance the deep learning processes of Large Language Models, fostering the development of sophisticated AI tools. In another project, we are currently researching the potential and challenges of Artificial Intelligence (AI) in automatic cataloguing.
Highlighting the challenges, the poster emphasizes the necessity of a community-organizing approach to address the multifaceted issues surrounding the modernisation of the authority record basic concept. Recognizing that reliable authority records demand a diverse range of expertise not solely present within libraries, the DNB advocates for active participation from various expert communities. Moreover, using authority records assumes that they are known, available and their benefits are understood. The poster illustrates pathways for collaborative engagement, demonstrating how an inclusive infrastructure facilitates the co-creation and enrichment of authority records. This includes a platform for the RDA DACH (https://sta.dnb.de/doc/RDA) rule set, that is participatory and a new visual web service for GND research, the GND Explorer (https://explore.gnd.network).
The research underscores that the success of AI applications is intricately tied to the quality and diversity of underlying data, emphasising the significance of standardization and harmonization through authority records. Ultimately, the poster asserts that the community-organizing approach is paramount in improving the abundance and richness of authority records and fostering widespread acceptance and utilisation within scientific communities, ensuring the gold standard for AI-based information management in libraries today and the future.
