1 |
Formalization of AMR Inference via Hybrid Logic Tableaux ...
|
|
|
|
Abstract:
AMR and its extensions have become popular in semantic representation due to their ease of annotation by non-experts, attention to the predicative core of sentences, and abstraction away from syntactic matter. An area where AMR and its extensions warrant improvement is formalization and suitability for inference, where it is lacking compared to other semantic representations, such as description logics, episodic logic, and discourse representation theory. This thesis presents a formalization of inference over a merging of Donatelli et al.’s (2018) AMR extension for tense and aspect with Pustejovsky et al.’s (2019) AMR extension for quantification and scope. Inference is modeled with a merging of Hansen’s (2007) tableau method for first-order hybrid logic with varying domain semantics (FHL) and Blackburn and Jørgensen’s (2012) tableau method for basic hybrid tense logic (BHTL). We motivate the merging of these AMR variants, present their interpretation and inference in the combination of FHL and BHTL, which ...
|
|
Keyword:
Abstract Meaning Representation; AMR; automated reasoning; Computational Linguistics; hybrid logic; Knowledge Representation; semantic representation; Semantics; tableau
|
|
URL: https://dx.doi.org/10.48617/etd.45 https://scholarworks.brandeis.edu/esploro/outputs/graduate/9924022111401921
|
|
BASE
|
|
Hide details
|
|
|
|