Broad-Coverage Semantic Parsing as Transduction

TL;DR

This paper introduces a unified neural transduction framework for broad-coverage semantic parsing, effectively improving state-of-the-art results across multiple semantic frameworks without pre-trained aligners.

Contribution

It proposes an attention-based neural transducer that unifies different semantic parsing tasks under a single transduction paradigm, enhancing performance and training efficiency.

Findings

Improves state-of-the-art on AMR and UCCA parsing tasks.

Achieves competitive results on SDP.

Effectively trains without pre-trained aligners.

Abstract

We unify different broad-coverage semantic parsing tasks under a transduction paradigm, and propose an attention-based neural framework that incrementally builds a meaning representation via a sequence of semantic relations. By leveraging multiple attention mechanisms, the transducer can be effectively trained without relying on a pre-trained aligner. Experiments conducted on three separate broad-coverage semantic parsing tasks -- AMR, SDP and UCCA -- demonstrate that our attention-based neural transducer improves the state of the art on both AMR and UCCA, and is competitive with the state of the art on SDP.