AMR Parsing via Graph-Sequence Iterative Inference

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 1290–1301 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 1290 AMR Parsing via Graph⇆Sequence Iterative Inference∗ Deng Cai The Chinese University of Hong Kong thisisjcykcd@gmail.com Wai Lam The Chinese University of Hong Kong wlam@se.cuhk.edu.hk Abstract We propose a new end-to-end model that treats AMR parsing as a series of dual decisions on the input sequence and the incrementally con- structed graph. At each time step, our model performs multiple rounds of attention, reason- ing, and composition that aim to answer two critical questions: (1) which part of the input sequence to abstract; and (2) where in the out- put graph to construct the new concept. We show that the answers to these two questions are mutually causalities. We design a model based on iterative inference that helps achieve better answers in both perspectives, leading to greatly improved parsing accuracy. Our ex- perimental results significantly outperform all previously reported SMATCH scores by large margins. Remarkably, without the help of any large-scale pre-trained language model (e.g., BERT), our model already surpasses previous state-of-the-art using BERT. With the help of BERT, we can push the state-of-the-art results to 80.2% on LDC2017T10 (AMR 2.0) and 75.4% on LDC2014T12 (AMR 1.0). 1 Introduction Abstract Meaning Representation (AMR) (Ba- narescu et al., 2013) is a broad-coverage semantic formalism that encodes the meaning of a sentence as a rooted, directed, and labeled graph, where nodes represent concepts and edges represent rela- tions (See an example in Figure 1). AMR parsing is the task of transforming natural language text into AMR. One biggest challenge of AMR parsing is the lack of explicit alignments between nodes (con- cepts) in the graph and words in the text. This char- acteristic not only poses great difficulty in concept ∗The work described in this paper is substantially sup- ported by grants from the Research Grant Council of the Hong Kong Special Administrative Region, China (Project Code: 14204418) and the Direct Grant of the Faculty of Engineering, CUHK (Project Code: 4055093). prediction but also brings a close tie for concept prediction and relation prediction. While most previous works rely on a pre-trained aligner to train a parser, some recent attempts in- clude: modeling the alignments as latent variables (Lyu and Titov, 2018), attention-based sequence- to-sequence transduction models (Barzdins and Gosko, 2016; Konstas et al., 2017; van Noord and Bos, 2017), and attention-based sequence-to-graph transduction models (Cai and Lam, 2019; Zhang et al., 2019b). Sequence-to-graph transduction models build a semantic graph incrementally via spanning one node at every step. This property is appealing in terms of both computational efficiency and cognitive modeling since it mimics what hu- man experts usually do, i.e., first grasping the core ideas then digging into more details (Banarescu et al., 2013; Cai and Lam, 2019). Unfortunately, the parsing accuracy of existing works including recent state-of-the-arts (Zhang et al., 2019a,b) remain unsatisfactory compared to human-level performance,1 especially in cases where the sentences are rather long …

Content truncated to 500 words