SGL: Speaking the Graph Languages of Semantic Parsing via Multilingual Translation

Abstract

Graph-based semantic parsing aims to represent textual meaning through directed graphs. As one of the most promising general-purpose meaning representations, these structures and their parsing have gained a significant interest momentum during recent years, with several diverse formalisms being proposed. Yet, owing to this very heterogeneity, most of the research effort has focused mainly on solutions specific to a given formalism. In this work, instead, we reframe semantic parsing towards multiple formalisms as Multilingual Neural Machine Translation (MNMT), and propose SGL, a many-to-many seq2seq architecture trained with an MNMT objective. Backed by several experiments, we show that this framework is indeed effective once the learning procedure is enhanced with large parallel corpora coming from Machine Translation: we report competitive performances on AMR and UCCA parsing, especially once paired with pre-trained architectures. Furthermore, we find that models trained under this configuration scale remarkably well to tasks such as cross-lingual AMR parsing: SGL outperforms all its competitors by a large margin without even explicitly seeing non-English to AMR examples at training time and, once these examples are included as well, sets an unprecedented state of the art in this task. We release our code and our models for research purposes at https: //github.com/SapienzaNLP/sgl.

BabelNet Meaning Representation: A Fully Semantic Formalism to Overcome Language Barriers

“…We just note that nothing prevents these notions from being incorporated into BMR as well. (Bevilacqua, Blloshmi, and Navigli 2021;Procopio, Tripodi, and Navigli 2021), which require large training sets. To make the BMR idea realistic, we put forward a procedure for creating a similar dataset for BMR.…”

Maximum Bayes Smatch Ensemble Distillation for AMR Parsing

“…Blloshmi et al (2020) show that one may not need alignment-based parsers for cross-lingual AMR, and model concept identification as a seq2seq problem. Procopio et al (2021) reframe semantic parsing as multilingual machine translation (MNMT) and propose a seq2seq architecture finetuned on pretrained-mBART with an MNMT objective. Cai et al (2021b) propose to use bilingual input to enable a model to predict more accurate AMR concepts.…”

“…Cross-lingual Parser Smatch scores on AMR2.0 human translated test sets. mBART mt ofProcopio et al (2021) indicates the mBART model fine-tuned on both semantic parsing tasks and the MT data. mBARTmmt ofCai et al (2021a) indicates an NMT model by(Tang et al, 2020), trained from mBART covering 50 languages.…”

ExtEnD: Extractive Entity Disambiguation

“…and Paşca (2006), ED aims to identify the actors involved in human language and, as such, has shown potential in downstream applications like Question Answering (Yin et al, 2016), Information Extraction (Ji and Grishman, 2011;Guo et al, 2013), Text Generation (Puduppully et al, 2019) and Semantic Parsing (Bevilacqua et al, 2021;Procopio et al, 2021).…”