Referentiedatabanken

Referentiedatabanken are databases designed to store and organize referents—entities in the real world or fictional domains—that are linked to linguistic expressions, texts, or other identifiers. A referent in such a database is typically given a unique referent ID and may be described by a set of attributes, including its type (for example person, place, organization, object), labels in multiple languages, and variants or spellings. The database also records contextual information such as geographic location, time period, and relations to other referents.

The primary purpose of referentiedatabanken is to support semantic annotation, information extraction, and linguistic research by providing a consistent, reusable reference system. They facilitate cross-corpus comparability, coreference resolution, and ontology development by linking words or phrases to standardized referents. In digital humanities and language documentation, such databases help preserve knowledge about named entities and other referents across languages and annotation schemes.

A typical data model includes core elements such as referent ID, preferred label, type or category, language, synonyms, and alternative spellings; contextual attributes like place, date, and domain; relations to other referents (for example part of, located in, related to); provenance and confidence scores; and links to external resources or ontologies (for instance Wikidata, domain-specific ontologies). Many referentiedatabanken support multilingual labeling and align with semantic web standards such as SKOS or RDF/OWL to enable interoperability.

Governance varies by institution but commonly involves curation policies, versioning, licensing, and update schedules. Access can be open or restricted, with attention to data quality, disambiguation processes, and privacy considerations when personal data are involved.