Overview
VerbNet (VN) is a large, freely available, online lexical resource for English verbs. It is structured as a hierarchical verb lexicon, linking syntactic and semantic patterns. VN extends Levin's verb classes by refining and adding subclasses to achieve coherence. Each verb class is described by thematic roles, selectional preferences, and frames. These frames detail syntactic descriptions and semantic representations, patterned on the Dynamic Event Model. VerbNet integrates with other resources like WordNet, PropBank, and FrameNet. Recent extensions have incorporated new classes from Korhonen and Briscoe (K&B) and Korhonen and Ryant (K&R), significantly expanding coverage of PropBank tokens and addressing limitations of Levin's original work. The resource is valuable for NLP tasks like semantic role labeling and parsing.