Semantic Inference at the Lexical-Syntactic Level

Abstract

Semantic inference is an important component in many natural language understanding applications. Classi-cal approaches to semantic inference rely on complex logical representations. However, practical applications usually adopt shallower lexical or lexical-syntactic rep-resentations, but lack a principled inference framework. We propose a generic semantic inference framework that operates directly on syntactic trees. New trees are inferred by applying entailment rules, which provide a unified representation for varying types of inferences. Rules were generated by manual and automatic meth-ods, covering generic linguistic structures as well as specific lexical-based inferences. Initial empirical eval-uation in a Relation Extraction setting supports the va-lidity of our approach.