About

SPRING (Symmetric ParsIng aNd Generation) is a state-of-the-art sequence-to-sequence model for parsing a sentence into an Abstract Meaning Representation (AMR) and, vice versa, generating a sentence from a given AMR graph, without many cumbersome external components.

SPRING model provided on this demonstration is trained on the concatenation of the AMR 3.0 corpus and the BioAMR, to offer a higher generalizability across different domains and genres.

This website provides:

• A Web interface for parsing a sentence into an AMR graph or generating a sentence given the AMR graph in a PENMAN serialized format.
• A RESTful API to programmatically perform AMR parsing or generation.
• A feedback system to suggest modifications of the SPRING output.
• Bibliography and model download section.

PUBLICATIONS

Michele Bevilacqua, Rexhina Blloshmi, Roberto Navigli. One SPRING to Rule Them Both: Symmetric AMR Semantic Parsing and Generation without a Complex Pipeline. Proceedings of the AAAI Conference on Artificial Intelligence (AAAI 2021), 35(14), 12564-12573.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the support of the ERC Consolidator Grant MOUSSE No. 726487, the ELEXIS project No. 731015 and the European Language Grid project No. 825627 (Universal Semantic Annotator, USeA) under the European Union's Horizon 2020 research and innovation programme.

LICENSE

SPRING is licensed under the CC BY-NC-SA 4.0 License.