Increasing CMDI’s Semantic Interoperability with schema.org

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2714–2720 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2714 Increasing CMDI’s Semantic Interoperability with schema.org Nino Meisinger, Thorsten Trippel, Claus Zinn University of Tuebingen Wilhelmstraße 19, 72074 Tuebingen, Germany {firstName.lastName}@uni-tuebingen.de Abstract The CLARIN Concept Registry (CCR) is the common semantic ground for most CMDI-based profiles to describe language- related resources in the CLARIN universe. While the CCR supports semantic interoperability within this universe, it does not extend beyond it. The flexibility of CMDI, however, allows users to use other term or concept registries when defining their metadata components. In this paper, we describe our use of schema.org, a light ontology used by many parties across disciplines. Keywords: Metadata Management, Semantic Interoperability, CMDI, schema.org 1. Introduction In practise, the semantic interoperability of CMDI- based metadata is based upon a single, CLARIN- based, registry. When metadata practitioners define CMDI profiles and their components, they usually ground the elementary metadata fields in the CLARIN concept registry (CCR). This helps applications that make use of CMDI-based metadata, such as the Vir- tual Language Observatory (VLO), to provide a se- mantic search across millions of resources described with many different CMDI profiles. The grounding of CMDI profiles with the CCR facilitates the map- ping of commonly used data categories to the search facets offered by the VLO. The overwhelming major- ity of data categories, however, are not discipline spe- cific, nor do they only make sense in the CLARIN con- text. In this paper, we argue that a grounding of CMDI- based profiles with vocabularies that are widely used across communities increases their semantic interoper- ability. Where possible, CMDI practitioners should not rely on community-specific (and hard to maintain) vo- cabularies. Instead they should opt for using vocab- ulary that is understood outside the CLARIN realm. To test our approach, we have developed a tool that transforms CMDI-based metadata based on CCR terms into CMDI-based metadata based on schema.org terms. Our TALAR repository will now export the traditional XML-based CMDI data together with a JSON-LD- based CMDI that makes use of the new terminology. 2. Background The Common Language Resources and Technology In- frastructure (CLARIN) enables research based on dig- ital language resources by offering advanced services to discover, explore, exploit, annotate, analyse, com- bine or archive language data, see clarin.eu. At the core of this infrastructure serves a metadata framework that makes it possible to describe language-related re- sources (data and tools) with a rich, expressive vocab- ulary. Expressive metadata is key to implementing the FAIR principles (Wilkinson et al., 2016) of findabil- ity of research data, accessibility, interoperability and reusability: metadata must use well-defined terms and value schemes that are shared within the research com- munity, and ideally understood across communities. Existing metadata schemes by themselves do not fill the bill. Bibliographic metadata schemes such as Dublin Core (dublincore.org/) or MARC21 (www. loc.gov/marc/bibliographic/) are standard- ised schemas from the library world. While they are …

Content truncated to 500 words