VerbNet

VerbNet is a large, publicly available lexicon of English verbs that organizes verbs into hierarchical classes based on shared syntactic behavior and semantic roles. It provides detailed, machine-readable descriptions of how verbs can be used in sentences, including subcategorization frames, thematic roles, and selectional restrictions. By grouping verbs that behave similarly, VerbNet supplies uniform analyses of verb syntax and meaning that support natural language processing and linguistic research.

Each class in VerbNet describes a set of verbs that share one or more frames. A class entry includes the syntactic frames in which its verbs typically appear (for example, transitive, ditransitive, or prepositional constructions), the thematic roles associated with the frames (such as agent, patient, theme, recipient, and goal), and example sentences. Members of a class inherit these frames and can be annotated with sense-specific information. VerbNet also links to related resources, including Levin-style classifications and cross-resource mappings to WordNet and PropBank-style roles.

VerbNet is widely used in natural language processing tasks, particularly semantic role labeling, shallow parsing, information extraction, and lexical semantics research. It provides a structured, testable representation of verb behavior that supports corpus analysis, machine translation, and the development of language technologies that require reliable verb semantics.

Originating in the lexical semantics research community, VerbNet has been developed and expanded since the early 2000s by researchers in computational linguistics. The resource continues to be maintained and updated to cover a broad portion of English verbs and to reflect regularities in verb syntax and argument structure.