High-Fidelity Lexical Axiom Construction from Verb Glosses

This paper presents a rule-based approach to constructing lexical axioms from WordNet verb entries in an expressive semantic representation, Episodic Logic (EL). EL differs from other representations in being syntactically close to natural language and covering phenomena such as generalized quantification, modification, and intensionality while still allowing highly effective inference. The presented approach uses a novel preprocessing technique to improve parsing precision of coordinators and incorporates frames, hand-tagged word senses, and examples from WordNet to achieve highly consistent semantic interpretations. EL allows the full content of glosses to be incorporated into the formal lexical axioms, without sacrificing interpretive accuracy, or verb-to-verb inference accuracy on a standard test set. Evaluation of semantic parser performance is based on EL-match, introduced here as a generalization of the smatch metric for semantic structure accuracy. On gloss parses, the approach achieves an ELmatch F1 score of 0.83, and a wholeaxiom F1 score of 0.45; verb entailment identification based on extracted axioms is competitive with the state-of-the-art.