Text generation from Abstract Meaning Representation graphs

Description

Abstract Meaning Representation (AMR; link) is a semantic representation language. An AMR is a single-rooted, directed graph representing the meaning of a sentence.

AMR-to-text generation is the task of converting AMR graphs into sentences. You can use this demo to test the best sequential, tree and graph encoders discussed in our paper.

We thank our funders: Bloomberg and EU H2020 (grant agreement: 688139, SUMMA).

The system’s source code can be found on github.

The multilingual dataset that we created can be found on LDC.

Paper

The system is described in detail in the paper that can be downloaded here.

@inproceedings{damonte2019gen,
  title={Structural Neural Encoders for AMR-to-text Generation},
  author={Damonte, Marco and Cohen, Shay B},
  booktitle={Proceedings of NAACL},
  year={2019}
}

Demo

Encoder:

AMR:




If you see an ERROR message, make sure your AMR is well-formed.

Sentence:

Graph: