realterms

Realterms is a term used in lexical semantics and natural language processing to describe lexical items that denote entities with verifiable existence in the external world. Realterms contrast with terms that refer to fictional, hypothetical, or abstract entities; for example, a realterm might be a proper noun such as Paris or Google, or a common noun that points to a concrete real-world referent like water when used in a specific, referential sense. The precise boundaries of what counts as a realterm vary between researchers, and the term is not universally standardized.

In practice, realterms are central to tasks such as named entity recognition, entity disambiguation, and knowledge-graph construction. Realterms are often linked to external knowledge bases (for instance Wikidata or DBpedia) to provide a stable identifier for a referent and to facilitate cross-domain linking. Challenges include homonymy, where a single surface form may refer to multiple real-world entities (Apple the company vs the fruit), and sense shifts over time as new entities emerge or old ones disappear.

Realterms can be categorized in several ways, including by referent type (person, location, organization, object, event) and by degree of referential commitment in discourse. Etymology is straightforward: the term is formed from 'real' plus 'terms', underscoring the link to actual referents.

See also: named entity, referent, proper noun, knowledge graph, entity linking, disambiguation.