Selectional Preferences Using a Lexical Hierarchy. UCAM investigated Bayesian selectional preference models that incorporate knowledge from a lexical hierarchy such as WordNet. These approaches are based either on “cutting” the hierarchy at an appropriate level of generalization or on a “walking” model that selects a path from the root to a leaf. The models were evaluated against human plausibility judgements and were shown to improve estimation of plausibility for out-of-vocabulary items.
Appears in 2 contracts
Sources: Grant Agreement, Grant Agreement