Dzung Phan, Vinicius Lima
INFORMS 2023
We proposed a framework and its implementation as a Python library for converting English utterances into higher-order logic (HOL) formulas. HOL extends first-order logic and provides flexibility for representing natural language semantics. Our library uses a broad-coverage and robust HPSG grammar for English to produce minimal recursive semantics (MRS) structures. These open-source technologies from the DELPH-IN Consortium balance a rigorous linguistic grounding and compositionality with practical aspects for natural language processing applications. Finally, we evaluated our approach over SICK, a popular dataset for text entailment.
Dzung Phan, Vinicius Lima
INFORMS 2023
Jehanzeb Mirza, Leonid Karlinsky, et al.
NeurIPS 2023
Hagen Soltau, Lidia Mangu, et al.
ASRU 2011
Liya Fan, Fa Zhang, et al.
JPDC