After clicking the following pictures, you can use the keyboard Tab key to switch to the previous, next and close buttons, or use the keyboard ← → key to switch to the previous and next picture, and use the keyboard Esc key to close the popup window

The National Central Library (NCL), in long-standing collaboration with the Library Association of the Republic of China (Taiwan), LAROC, routinely hosts the IO Talk forum to facilitate professional exchange and academic exploration regarding emerging topics in information organization. On June 16, 2026, the forum convened under the theme "Ethical Challenges for Knowledge Organization in the Age of Artificial Intelligence,” inviting Dr. Natalia Bolfarini Tognoli from the Department of Information Science, Fluminense Federal University in Brazil, as the keynote speaker to share her scholarly insights into the ethical concerns of knowledge organization. The event was expertly moderated by Professor Ya-Ning Chen and Professor Wei-Ning Cheng from the Graduate Institute of Library and Information Studies at National Taiwan Normal University.

The moderator Professor Ya-Ning Chen introduced the keynote speaker, Dr. Tognoli, who serves concurrently the Editor-in-Chief of the SSCI international journal<<Knowledge Organization>> and a distinguished expert in the ethics studies of knowledge organization. Professor Chen noted that the theme represents an emerging issue that has been seldom explored within Taiwan’s LIS field, and encouraged attendees to actively engage in scholarly exchange with Dr. Tognoli.

In her presentation, Dr. Tognoli shared her academic background, which primarily focuses on archival theory and knowledge organization from a social justice perspective. While clarifying that she is not an artificial intelligence expert per se, Dr. Tognoli demonstrated how the ethical challenges of the AI era can be critically examined through the lens of knowledge organization. Researching the ethics of knowledge organization, Dr. Tognoli emphasized, entails analyzing the underlying power dynamics and subjective choices: Who possesses the authority to describe, represent, and evaluate knowledge? Which cultures and perspectives are reflected within knowledge organization system—and in what manner? Conversely, which communities are marginalized or excluded from these frameworks, thereby being effectively "silenced" within mainstream discourse?

Although artificial intelligence operations rely inherently on data, raw data must be human-curated and structured to generate datasets and metadata. Furthermore, semantic precision must be enhanced through specialized conceptual frameworks such as ontologies, taxonomies, and knowledge graphs. This structuring enables AI systems to execute computations and deliver responses that fulfill user requests, thereby underscoring the profound significance of knowledge organization to artificial intelligence. Grappling the dilemma of information explosion, contemporary AI depends increasingly on infrastructures capable of structuring information. Raw data must be organized to become knowledge, rendering information objects meaningful, representable, retrievable, and fully exploitable.

Dr. Tognoli emphasized that knowledge organization is by no means objective or neutral. Tools and systems engineered under the guidance of information professionals inevitably carry the perspectives and biases of dominant groups (such as western powers or heterosexual Caucasian males). Meanwhile, the voices of minority groups (including females, indigenous peoples, and LGBTQIA+ communities) are often obscured or suppressed that significantly impacts users' information retrieval, access, and epistemology, to the distinct disadvantage of the marginalized populations. Consequently, AI systems operating upon biased knowledge organization frameworks will inherently perpetuate these biases across three interconnected levels: Data level: training data derived from specific temporal-spatial contexts or dominant demographics leads to data biases; conceptual level: specific classifications and conceptual frameworks used to represent data subjectively shape the meaning of that information; algorithmic level: algorithmic decision-making in information systems reinforces specific classifications, thereby exacerbating inequalities and systemic biases.  These three forms of epistemic bias are deeply intertwined, creating a vicious cycle that amplifies bias throughout the system.

Regarding the application of knowledge organization in AI, ontologies, as formal representations of concepts and relationships within specific domains, serve as a vital bridge connecting knowledge organization and AI, supporting reasoning and inference within AI systems by organizing domain-specific concepts. knowledge graphs, as structured representations of real-world entities and relationships, link entities into a visualized semantic web. This infrastructure facilitates the comprehension of connections between disparate information fragments, thereby supporting complex queries across multiple data sources, enhancing AI applications, and optimizing information retrieval and deduction capabilities.

Concluding her lecture, Dr. Tognoli asserted that artificial intelligence is actively reshaping the ecosystem of the library profession and increasingly intervening in the processes of knowledge organization and information retrieval. Far from being replaced, the role of LIS researchers and practitioners has become more pivotal than ever. Today, in addition to mastering technical competencies, professionals must engage in critical reflection regarding knowledge representation, classification schemes, transparency, and social responsibility to effectively address the ethical challenges of the AI era.

The subsequent Q&A and discussion was moderated by Professor Wei-Ning Cheng, during which attendees of the LIS field enthusiastically raised questions and sought advice from the speaker. Dr. Tognoli offered the following recommendations for professionals in the field: Collaborate with experts across diverse disciplines to co-create ethical and robust AI tools; actively listen to the public and construct knowledge organization systems and tools rooted in community cultures; advance specialized domain knowledge to enhance awareness of domain-specific biases, thereby mitigating the risk of introducing prejudice into these systems.

The event concluded successfully amidst vibrant academic exchange among the keynote speaker, moderators, and participants. The National Central Library extends its sincere gratitude to Dr. Tognoli for her illuminating presentation, to the two moderators for their insightful guidance, and to the 61 Taiwanese LIS professionals whose enthusiastic attendance contributed to the success of the event.