Factorising AMR generation through syntax
This addresses the challenge of text generation from semantic graphs for natural language processing applications, offering an incremental improvement by explicitly handling underspecification.
The paper tackles the underspecified problem of generating text from Abstract Meaning Representation (AMR) by decomposing it into two steps: generating a syntactic structure first, then the surface form, achieving state-of-the-art single model performance without extra unlabelled data and producing human-judged meaning-preserving syntactic paraphrases.
Generating from Abstract Meaning Representation (AMR) is an underspecified problem, as many syntactic decisions are not constrained by the semantic graph. To explicitly account for this underspecification, we break down generating from AMR into two steps: first generate a syntactic structure, and then generate the surface form. We show that decomposing the generation process this way leads to state-of-the-art single model performance generating from AMR without additional unlabelled data. We also demonstrate that we can generate meaning-preserving syntactic paraphrases of the same AMR graph, as judged by humans.