Natural Language Inference as Triggered Submodel Search

Daniel Hardt, Copenhagen Business School

I propose a simple, general framework for the interaction of inference with natural language interpretation. First, inference is only available when triggered by the violation of highly ranked constraints. Second, inference is constrained to be a search for a minimal submodel. I show that this correctly captures facts concerning deaccenting, ellipsis, and reciprocal interpretation.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.