Neural Semantic Parsing by Character-based Translation: Experiments with Abstract Meaning Representations

Abstract

We evaluate the character-level translation method for neural semantic parsing on a large corpus of sentences annotated with Abstract Meaning Representations (AMRs). Using a sequence-to-sequence model, and some trivial preprocessing and postprocessing of AMRs, we obtain a baseline accuracy of 53.1 (F-score on AMR-triples). We examine five different approaches to improve this baseline result: (i) reordering AMR branches to match the word order of the input sentence increases performance to 58.3; (ii) adding part-of-speech tags (automatically produced) to the input shows improvement as well (57.2); (iii) So does the introduction of super characters (conflating frequent sequences of characters to a single character), reaching 57.4; (iv) optimizing the training process by using pre-training and averaging a set of models increases performance to 58.7; (v) adding silver-standard training data obtained by an off-the-shelf parser yields the biggest improvement, resulting in an F-score of 64.0. Combining all five techniques leads to an F-score of 71.0 on holdout data, which is state-of-the-art in AMR parsing. This is remarkable because of the relative simplicity of the approach.

Camera ready for CLIN 2017 journal

Cited by in corpus (25)

• OpenNMT: Neural Machine Translation Toolkit
• AMR Dependency Parsing with a Typed Semantic Algebra
• Guided Neural Language Generation for Abstractive Summarization using Abstract Meaning Representation
• Rewarding Smatch: Transition-Based AMR Parsing with Reinforcement Learning
• AMR Parsing via Graph-Sequence Iterative Inference
• AMR Parsing as Sequence-to-Graph Transduction
• Improving AMR Parsing with Sequence-to-Sequence Pre-training
• Broad-Coverage Semantic Parsing as Transduction
• AMR Parsing with Action-Pointer Transformer
• Core Semantic First: A Top-down Approach for AMR Parsing
• Separating Argument Structure from Logical Structure in AMR
• AMR Normalization for Fairer Evaluation
• Automatic Accuracy Prediction for AMR Parsing
• Exploring Neural Methods for Parsing Discourse Representation Structures
• The First Shared Task on Discourse Representation Structure Parsing
• Transition-Based Generation from Abstract Meaning Representations
• SemBleu: A Robust Metric for AMR Parsing Evaluation
• Structure-aware Fine-tuning of Sequence-to-sequence Transformers for Transition-based AMR Parsing
• Making Better Use of Bilingual Information for Cross-Lingual AMR Parsing
• Multilingual AMR Parsing with Noisy Knowledge Distillation
• Ensembling Graph Predictions for AMR Parsing
• A Differentiable Relaxation of Graph Segmentation and Alignment for AMR Parsing
• Character-level Representations Improve DRS-based Semantic Parsing Even in the Age of BERT
• An empirical evaluation of AMR parsing for legal documents
• The problem with probabilistic DAG automata for semantic graphs