Orfeas Menis Mastromichalakis, Jason Liartis, Kristina Rose, Antoine Isaac, Giorgos Stamou. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Europeana Translate is a project funded under the Connecting European Facility with the objective to take advantage of state-of-the-art machine translation in order to increase the multilinguality of resources in the cultural heritage domain

Cultural heritage institutions hold collections of printed newspapers that are valuable resources for the study of history, linguistics and other Digital Humanities scientific domains. Effective retrieval of newspapers content based on metadata only is a task nearly impossible, making the retrieval based on (digitized) full-text particularly relevant. Europeana, Europe’s Digital Library, is in the position to provide access to large newspapers collections with full-text resources. Full-text corpora are also relevant for Europeana’s objective of promoting the usage of cultural heritage resources for use within research infrastructures. We have derived requirements for aggregating and publishing Europeana’s newspapers full-text corpus in an interoperable way, based on investigations into the specific characteristics of cultural data, the needs of two research infrastructures (CLARIN and EUDAT) and the practices being promoted in the International Image Interoperability Framework (IIIF) community. We have then defined a "full-text profile" for the Europeana Data Model, which is being applied to Europeana’s newspaper corpus.

A very large number of online cultural heritage (CH) resources is made available through numerous digital libraries. To address the difficulties of discoverability in CH, the common practice is metadata aggregation, where centralized efforts like Europeana facilitate discoverability by collecting the resources' metadata. In the last years, the CH domain has invested in data models for Linked Data (LD) representation of CH metadata. LD, however, also has potential for innovating metadata aggregation. We present the results of a pilot case study within the Europeana Network. In this pilot, the National Library of The Netherlands plays the role of initial data provider, with the Dutch Digital Heritage Network the one of intermediary service providing datasets to Europeana. We analysed the requirements for an LD aggregation solution and defined a workflow that fulfils the same functional requirements as Europeana's current solution. The workflow was put into practice within the pilot and led to the development of several software components for managing datasets, harvesting LD, data analysis and integration. Our analysis of the experience discusses the effort of adopting such an LD approach for data providers and aggregators, the expertise required by CH data analysts, and the supporting tools required for semantic data.

Simple Knowledge Organization System (SKOS) provides a data model and vocabulary for expressing Knowledge Organization Systems (KOSs) such as thesauri and classification schemes in Semantic Web applications. This paper presents the main components of SKOS and their formal expression in Web Ontology Language (OWL), providing an extensive account of the design decisions taken by the Semantic Web Deployment (SWD) Working Group of the World Wide Web Consortium (W3C), which between 2006 and 2009 brought SKOS to the status of W3C Recommendation. The paper explains key design principles such as “minimal ontological commitment” and systematically cites the requirements and issues that influenced the design of SKOS components. By reconstructing the discussion around alternative features and design options and presenting the rationale for design decisions, the paper aims at providing insight into how SKOS turned out as it did, and why. Assuming that SKOS, like any other successful technology, may eventually be subject to revision and improvement, the critical account offered here may help future editors approach such a task with deeper understanding.