AMR Parsing as Graph Prediction with Latent Alignment

Chunchuan Lyu, Ivan Titov

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Abstract meaning representations (AMRs) are broad-coverage sentence-level semantic representations. AMRs represent sentences as rooted labeled directed acyclic graphs. AMR parsing is challenging partly due to the lack of annotated alignments between nodes in the graphs and words in the corresponding sentences. We introduce a neural parser which treats alignments as latent variables within a joint probabilistic model of concepts, relations and alignments. As exact inference requires marginalizing over alignments and is infeasible, we use the variational auto-encoding framework and a continuous relaxation of the discrete alignments. We show that joint modeling is preferable to using a pipeline of align and parse. The parser achieves the best reported results on the standard benchmark (73.6% on LDC2016E25).
Original languageEnglish
Title of host publicationProceedings of the 56th Annual Conference of the Association for Computational Linguistics (ACL)
Place of PublicationMelbourne, Australia
PublisherAssociation for Computational Linguistics
Pages397-407
Number of pages11
Publication statusPublished - Jul 2018
Event56th Annual Meeting of the Association for Computational Linguistics - Melbourne Convention and Exhibition Centre, Melbourne, Australia
Duration: 15 Jul 201820 Jul 2018
http://acl2018.org/

Conference

Conference56th Annual Meeting of the Association for Computational Linguistics
Abbreviated titleACL 2018
Country/TerritoryAustralia
CityMelbourne
Period15/07/1820/07/18
Internet address

Fingerprint

Dive into the research topics of 'AMR Parsing as Graph Prediction with Latent Alignment'. Together they form a unique fingerprint.

Cite this