Wordnets

WordNets are lexical databases that organize words into structured networks of meaning, typically through sets of cognitive synonyms called synsets that express a single concept. The most widely used example is WordNet, a large English resource developed at Princeton University, but the term also encompasses a family of WordNet-style resources created for many other languages. These multilingual and cross-lingual WordNets enable linking translations and shared concepts across languages.

In their common form, WordNets arrange synsets with short definitions (glosses) and example sentences. Words are linked to one or more synsets, and synsets are connected by semantic relations such as hypernymy and hyponymy (is-a relationships), meronymy and holonymy (part-whole), antonymy, and others. This structure supports tasks that rely on semantic similarity, word sense disambiguation, and lexical inference. Extensions may include domain labels, emotion or affect annotations, and links to external resources.

WordNets originated with WordNet at Princeton in the 1980s and inspired a broader ecosystem of multilingual projects. Notable initiatives include EuroWordNet, which integrated several European languages, and the Open Multilingual WordNet, which provides cross-language connections and downloadable data. Other language-specific WordNets extend the core concept to reflect language-specific senses and usages.

Applications of WordNets include natural language processing, information retrieval, machine translation, question answering, and semantic search. Limitations include coverage gaps for less common words or domain-specific vocabularies, evolving language use, and the manual effort required to maintain and align cross-language links.