Transition-based Semantic Dependency Parsing with Pointer Networks

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7035–7046 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 7035 Transition-based Semantic Dependency Parsing with Pointer Networks Daniel Fern´andez-Gonz´alez and Carlos G´omez-Rodr´ıguez Universidade da Coru˜na, CITIC FASTPARSE Lab, LyS Group Depto. de Ciencias de la Computaci´on y Tecnolog´ıas de la Informaci´on Campus de Elvi˜na, s/n, 15071 A Coru˜na, Spain d.fgonzalez@udc.es, carlos.gomez@udc.es Abstract Transition-based parsers implemented with Pointer Networks have become the new state of the art in dependency parsing, excelling in producing labelled syntactic trees and out- performing graph-based models in this task. In order to further test the capabilities of these powerful neural networks on a harder NLP problem, we propose a transition system that, thanks to Pointer Networks, can straight- forwardly produce labelled directed acyclic graphs and perform semantic dependency pars- ing. In addition, we enhance our approach with deep contextualized word embeddings extracted from BERT. The resulting system not only outperforms all existing transition- based models, but also matches the best fully- supervised accuracy to date on the SemEval 2015 Task 18 English datasets among previous state-of-the-art graph-based parsers. 1 Introduction In dependency parsing, the syntactic structure of a sentence is represented by means of a labelled tree, where each word is forced to be attached exclu- sively to another that acts as its head. In contrast, semantic dependency parsing (SDP) (Oepen et al., 2014) aims to represent binary predicate-argument relations between words of a sentence, which re- quires producing a labelled directed acyclic graph (DAG): not only semantic predicates can have mul- tiple or zero arguments, but words from the sen- tence can be attached as arguments to more than one head word (predicate), or they can be outside the SDP graph (being neither a predicate nor an argument) as shown in the examples in Figure 1. Since existing dependency parsers cannot be di- rectly applied, most SDP research has focused on adapting them to deal with the absence of single- head and connectedness constraints and to produce an SDP graph instead. Figure 1: Sentence from the SemEval 2015 Task 18 de- velopment set parsed with semantic dependencies fol- lowing the DM, PAS and PSD formalisms. As in dependency parsing, we can find two main families of approaches to efficiently generate accu- rate SDP graphs. On the one hand, graph-based algorithms have drawn more attention since adapt- ing them to this task is relatively straightforward. In particular, these globally optimized methods in- dependently score arcs (or sets of them) and then search for a high-scoring graph by combining these scores. From one of the first graph-based DAG parsers proposed by McDonald and Pereira (2006) to the current state-of-the-art models (Wang et al., 2019; He and Choi, 2019), different graph-based SDP approaches have been presented, providing ac- curacies above their main competitors: transition- based DAG algorithms. A transition-based parser generates a sequence of actions to incrementally build a valid graph (usu- ally from left to right). This is typically done by local, …

Content truncated to 500 words