Broad-coverage CCG Semantic Parsing with AMR

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1699–1710, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Broad-coverage CCG Semantic Parsing with AMR Yoav Artzi∗ Dept. of Computer Science and Cornell Tech Cornell University New York, NY 10011 yoav@cs.cornell.edu Kenton Lee Luke Zettlemoyer Computer Science & Engineering University of Washington Seattle, WA 98195 {kentonl, lsz}@cs.washington.edu Abstract We propose a grammar induction tech- nique for AMR semantic parsing. While previous grammar induction techniques were designed to re-learn a new parser for each target application, the recently anno- tated AMR Bank provides a unique op- portunity to induce a single model for un- derstanding broad-coverage newswire text and support a wide range of applications. We present a new model that combines CCG parsing to recover compositional aspects of meaning and a factor graph to model non-compositional phenomena, such as anaphoric dependencies. Our ap- proach achieves 66.2 Smatch F1 score on the AMR bank, significantly outperform- ing the previous state of the art. 1 Introduction Semantic parsers map sentences to formal repre- sentations of their meaning (Zelle and Mooney, 1996; Zettlemoyer and Collins, 2005; Liang et al., 2011). Existing learning algorithms have primar- ily focused on building actionable meaning repre- sentations which can, for example, directly query a database (Liang et al., 2011; Kwiatkowski et al., 2013) or instruct a robotic agent (Chen, 2012; Artzi and Zettlemoyer, 2013b). However, due to their end-to-end nature, such models must be re- learned for each new target application and have only been used to parse restricted styles of text, such as questions and imperatives. Recently, AMR (Banarescu et al., 2013) was proposed as a general-purpose meaning represen- tation language for broad-coverage text, and work is ongoing to study its use for variety of appli- cations such as machine translation (Jones et al., 2012) and summarization (Liu et al., 2015). The ∗Work done at the University of Washington. AMR meaning bank provides a large new corpus that, for the first time, enables us to study the problem of grammar induction for broad-coverage semantic parsing. However, it also presents sig- nificant challenges for existing algorithms, in- cluding much longer sentences, more complex syntactic phenomena and increased use of non- compositional semantics, such as within-sentence coreference. In this paper, we introduce a new, scalable Combinatory Categorial Grammar (CCG; Steedman, 1996, 2000) induction approach that solves these challenges with a learned joint model of both compositional and non-compositional se- mantics, and achieves state-of-the-art performance on AMR Bank parsing. We map sentences to AMR structures in a two- stage process (Section 5). First, we use CCG to construct lambda-calculus representations of the compositional aspects of AMR. CCG is designed to capture a wide range of linguistic phenomena, such as coordination and long-distance dependen- cies, and has been used extensively for semantic parsing. To use CCG for AMR parsing we define a simple encoding for AMRs in lambda calculus, for example, as seen with the logical form z and AMR a in Figure 1 for …

Content truncated to 500 words