Papers: Cultural Heritage & Digital Humanities
- Starts at
- Wed, Aug 5, 2026, 10:30 KST
- Finishes at
- Wed, Aug 5, 2026, 12:30 KST
- Venue
- Room A
Presentations
Multilingual Metadata: Aligning Digital Heritage Systems with Cultural Values
Authors: Robin Dresel, Pamela Low
Metadata systems traditionally prioritise technical efficiency over cultural authenticity. This paper examines how Singapore's National Library Board redesigned metadata practices for two major cultural heritage digital projects—the Encyclopedia of Singapore Tamils (EST) and Prominent Malays of Singapore (PMoS)—to align digital systems with community values.
Breaking from locally established conventions, we implemented largely monolingual metadata records, creating separate collection identifiers, language-specific navigation paths, and culturally-aware controlled vocabularies. This approach required overcoming technical constraints in content management systems designed for English-language dominant workflows.
Our methodology involved close collaboration with over 500 community contributors. Key innovations include collection name separation, multilingual controlled vocabulary integration, and community-driven category translation that reflects cultural mental models rather than literal translations.
Implementation demonstrates that separate monolingual metadata records can preserve cultural authenticity while maintaining system functionality. These approaches create valuable training data for AI systems working with multilingual cultural heritage, offering a replicable model for institutions seeking to align digital systems with diverse community needs rather than technical convenience.
-
Robin Dresel
Assistant Director / Senior Librarian
National Library Board Singapore
Robin Dresel is Assistant Director, Metadata Services at the National Library Board Singapore, managing teams responsible for digital resources and non-purchase collections such as legal deposit, rare items, and donations. Drawing on over 20 years in libraries, he works at the intersection of cataloguing operations and technology, with a growing interest in how AI and system design can better serve diverse communities and collections. Recent studies in Digital Humanities sparked his curiosity about how humans and systems interact, and what that means for metadata practice.
TEI Encoding as Infrastructure for Meaning-Driven AI in Portuguese Literature
Authors: Diego Emanuel Giménez Celano
This presentation explores how TEI encoding can serve as infrastructure for meaning‑driven AI in Portuguese literature. Drawing on the encoding of Luís de Camões’ Os Lusíadas, it shows how metadata functions as epistemic governance, shaping what AI systems can recognize and reproduce as knowledge. Rather than treating AI as an interpretative agent, the talk frames it within a semantic architecture designed by human editors. By highlighting ethical implications and practical workflows, it argues that responsible AI participation in cultural knowledge depends on the design of metadata environments, not just algorithmic correction.
-
Diego Emanuel Giménez Celano
Assistant Professor
University of Macau
Diego Emanuel Giménez Celano is Professor of Portuguese Literature at the University of Macau. He holds a PhD in Literature and Thought from the University of Barcelona, with a dissertation on Fernando Pessoa’s The Book of Disquiet. He was a fellow of the Calouste Gulbenkian Foundation and a researcher on "No Problem Has a Solution: A Digital Archive of the Book of Disquiet" at the University of Coimbra. He was a postdoctoral researcher at the State University of Londrina, collaborates with Camões Lab, and is PI of "Portuguese Literary Studies: Texts, Readings, and Digital Approaches".
From MARC to Linked Open Data: AI-Driven Entity Extraction from Hebrew Manuscript Metadata Using Distant Supervision
Authors: Alexander Goldberg, Gila Prebor, Avshalom Elmalech
Cultural heritage institutions preserve invaluable provenance information in Machine-Readable Cataloging (MARC) records, yet much of this knowledge remains trapped in unstructured note fields, inaccessible to computational analysis. Transforming these legacy catalogs into Linked Open Data (LOD) requires extracting structured person-role relationships—identifying authors, scribes, owners, and censors—from cataloger narratives. This study presents an AI-driven system that uses distant supervision from MARC metadata itself to automatically generate training data, eliminating the prohibitive cost of manual annotation for specialized cultural heritage domains. By exploiting the dual structure of catalog records, where structured fields provide authoritative labels and unstructured notes provide context, we achieve 85.70% F1 for person extraction and 100% role classification accuracy, outperforming general Hebrew NER models by +55.55% F1. Our approach demonstrates how existing metadata can be leveraged to train AI systems that align with the values of cultural heritage preservation: accuracy, provenance tracking, and semantic enrichment. The extracted entities populate ontology instances based on CIDOC-CRM and IFLA-LRM, enabling computational analysis of scribal networks and manuscript circulation at scale.
-
Avshalom Elmalech
Researcher
Bar-Ilan University
Avshalom Elmalech is a researcher at Bar-Ilan University with a PhD in Computer Science, working at the intersection of applied artificial intelligence and digital humanities. His research bridges information science and AI by examining how deep learning methods can be effectively applied to humanities data. He has contributed practical frameworks for guiding digital humanities scholars in choosing and adapting NLP and deep learning approaches under constraints such as limited training data and domain specificity. -
Gila Prebor
Associate Professor, Department of Information Science
Bar-Ilan University
Gila Prebor is a researcher in the Department of Information Science at Bar-Ilan University, specializing in Hebrew manuscripts, paleography, and knowledge organization. Her research combines codicology and bibliography with Digital Humanities, focusing on AI, Handwritten Text Recognition (HTR), and Semantic Web technologies for cultural heritage data. She is co-editor of Alei Sefer and has received grants from the ISF, the EU, and Israel’s Ministry of Innovation, Science, and Technology. Her recent work applies Linked Data and AI to the study of Hebrew manuscripts.
Decolonizing Metadata: Lessons from Stolen Relations’ Controlled Vocabulary Development
Authors: Mairelys Lemus-Rojas, Patrick Rashleigh, Khanh Vo
Metadata should be understood as an interpretive practice and not just as a technical framework for describing digital objects. It holds power and facilitates community engagement. Within the digital humanities arena, metadata plays a pivotal role in narrating and recovering stories that have remained obscured or misrepresented in historical records. This raises a fundamental question: how can we more humanly describe Indigenous communities whose identities and relationships to kinship and culture have been misrepresented in colonial records? This paper examines the role of controlled vocabularies in shaping the representation of Indigenous histories in Stolen Relations’ digital humanities project, positioning metadata as a form of archival intervention. It demonstrates how iterative feedback informs the refinement, implementation, or creation of controlled vocabularies and positions metadata as a space where descriptive practices are examined and reshaped.
-
Mairelys Lemus-Rojas
Head of Digital Scholarship
University of Central Florida
Mairelys Lemus-Rojas is the Head of Digital Scholarship at the University of Central Florida Libraries. She oversees Digital Initiatives, Open Scholarship, and the Digital Exploration Center, a digital scholarship hub to learn, engage, and collaborate on digital projects. Previously, she worked as the Head of Open Metadata Production and Initiatives at Brown University. As a strong advocate for open knowledge and an active contributor to Wikimedia projects, Mairelys is committed to democratizing access to information by amplifying the visibility of underrepresented communities.
Equitable Metadata for Diverse Voices: Sustainable Computational Poetry Analysis with HathiTrust Extracted Features
Authors: Kahyun Choi, You Peng, Gyuri Kang
The retirement of the HathiTrust Research Center (HTRC) infrastructure raises questions about continuing computational research on in-copyright collections in the HathiTrust Digital Library (HTDL). Since HTRC has actively supported inclusive research on underrepresented groups, the HTDL collections serve as a crucial test case for exploring post-HTRC workflows. To address this, we share an augmented dataset of American poetry by poets from historically underrepresented groups in the HTDL. Mapping this collection to HTRC Extracted Features (EF) v2.5 demonstrated that EF is highly reliable with high retrieval coverage, achieving a 100\% match. Our computational linguistic analysis shows that EF effectively captures group-specific diversity, such as non-standard English, indigenous languages, and multilingual vocabularies. These findings indicate that adapting ML and NLP tools to properly handle such linguistic variation is essential to mitigate bias and marginalization. Although the lack of full-text limits structural analysis, EF remains a sustainable and highly useful resource for word-based research well beyond the HTRC's retirement.
-
Kahyun Choi
Assistant Professor
University of Illinois Urbana-Champaign
Kahyun Choi is an Assistant Professor in the School of Information Sciences at the University of Illinois Urbana-Champaign. She received her PhD from UIUC. Her research applies computational methods and machine learning to cultural data, focusing on computational analysis of poetry and music, human–AI co-creation of cultural metadata, and ethical AI for digital libraries. Her work spans Music Information Retrieval and Digital Humanities. She received the 2022 IMLS Early Career Grant and the 2021 IMLS National Leadership Grant, and the 2023 IU Trustees Teaching Award.
Metadata-Driven Semantic Interoperability: The HerStory-NeSyAI project for Trustworthy Neurosymbolic AI in Digital Humanities
Authors: Miquel Centelles Velilla, Matheus Jenevain, Elena Gómez, Núria Ferran-Ferrer
This short paper presents HerStory-NeSyAI as a work-in-progress neurosymbolic project in Digital Humanities. The project addresses fragmented historical datasets through metadata-driven semantic interoperability, combining a knowledge graph, ontology layer, and retrieval-augmented generation (RAG). It treats interoperability as a condition for epistemic justice and explores how graph-grounded metadata can improve transparency, traceability, and data integrity while mitigating bias, hallucinations, and poisoning risks. In operational terms, these goals are implemented through a prototype, which enables a three-path strategy for connecting heterogeneous datasets.
-
Miquel Centelles Velilla
Universitat de Barcelona
Professor at the University of Barcelona in Information Science, with prior experience at Universitat Pompeu Fabra’s library as coordinator and library assistant. Trained in Library & Information Science and Linguistics, with doctoral studies in cognitive science and language. Teaching and research focus on digital content and knowledge organization, including EPUB3 e-book metadata, RDF/linked data, taxonomies, and accessible multimodal learning resources, supported by publications, projects, and conference contributions. Currently working on neurosymbolic AI using knowledge graphs.
FAIR Open Metadata: A Case Study of RePEc
Authors: Anna Oates Schlaack, Christian Zimmermann
This report describes a research project about the use of RePEc metadata and its accordance with FAIR principles. The authors summarize successful aspects of a metadata schema that has underpinned the research landscape in economics for nearly thirty years. This report describes the use cases of 100 studies that have leveraged RePEc metadata and future research that will document the challenges of metadata citation and opportunities to engender open metadata efforts.
-
Anna Oates Schlaack
Assistant Professor, Cataloging & Metadata Librarian
University of Illinois Urbana-Champaign
Anna Oates Schlaack (she/her) is an Assistant Professor and the Cataloging and Metadata Librarian at the University of Illinois Urbana-Champaign, where she leads special formats and English-language monographic cataloging. While using the Modernist Journals Project as a student in the humanities, she found her calling–curating and describing information so that it can be used in novel ways. Whether to support digital humanities research, personal genealogical research, or scientometrics, Schlaack is committed to stewarding open metadata and information resources for use across the globe.
An Exploratory Study on Genre Labeling of Online Comic Reading Platforms in Taiwan
Authors: Tzu-Yun Chien, Li-Min Huang
As online comic reading continues to grow in Taiwan, online platforms have become important sites for both comic consumption and discovery. This exploratory study examines genre-labeling practices on three major online comic platforms in Taiwan. We collected genre terms from the platforms’ Traditional Chinese and English interfaces and generated a list of 35 unique English genre terms. These terms were then mapped onto a facet framework drawn from previous literature. Our preliminary findings show that platform genre labels function as multidimensional access points rather than simple genre categories, representing aspects such as setting, mood, plot or narrative, and production context. Cross-platform differences in labeling granularity and cross-language differences in semantic scope were also observed. The findings may inform the development of more consistent and user-friendly genre labels for digital comic environments.
-
Tzu-Yun Chien
National Taiwan University
Tzu-Yun Chien is a Master’s student in the Department of Library and Information Science at National Taiwan University. Her research interest focusing on human–computer interaction and information behavior. Her prior research on user behaviors in AI-assisted tasks has been published as a full conference paper. Her ongoing master's thesis focuses on the differences between existing genre categorization frameworks, platform labeling practices, and user interpretations within online comics.
Hosted by
