Getting the Most out of AMR Parsing

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 1257–1268 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Getting the Most out of AMR Parsing Chuan Wang and Nianwen Xue Brandeis University {cwang24,xuen}@brandeis.edu Abstract This paper proposes to tackle the AMR parsing bottleneck by improving two com- ponents of an AMR parser: concept iden- tification and alignment. We first build a Bidirectional LSTM based concept iden- tifier that is able to incorporate richer contextual information to learn sparse AMR concept labels. We then extend an HMM-based word-to-concept alignment model with graph distance distortion and a rescoring method during decoding to in- corporate the structural information in the AMR graph. We show integrating the two components into an existing AMR parser results in consistently better performance over the state of the art on various datasets. 1 Introduction Abstract Meaning Representation (AMR) (Ba- narescu et al., 2013) is a semantic representation where the meaning of a sentence is encoded as a rooted, directed graph. A number of AMR parsers have been developed in recent years (Flanigan et al., 2014; Wang et al., 2015b; Artzi et al., 2015; Pust et al., 2015; Peng et al., 2015; Zhou et al., 2016; Goodman et al., 2016a), and the initial benefit of AMR parsing has been demon- strated in various downstream applications such as Information Extraction (Pan et al., 2015; Huang et al., 2016), Machine Comprehension (Sachan and Xing, 2016), and Language Generation (Flani- gan et al., 2016b; Butler, 2016). However, AMR parsing parsing accuracy is still in the high 60%, as measured by the SMatch score (Cai and Knight, 2013), and a significant improvement is needed in order for it to positively impact a larger number of applications. Previous research has shown that concept iden- tification is the bottleneck to further improvement of AMR parsing. For example, JAMR (Flani- gan et al., 2014), the first AMR parser, is able to achieve an F-score of 80% (close to the inter- annotator agreement of 83) if gold concepts are provided. Its parsing accuracy drops sharply to 62.3% when the concepts are identified automat- ically. One of the challenges in AMR concept iden- tification is data sparsity. A large portion of AMR’s concepts are either word lemmas or sense-disambiguated lemmas drawn from Prop- bank (Palmer et al., 2005). Since the AMR Bank is relatively small, many of the concept labels in the development or test set only occur a few times or never appear in the training set. Werling et al. (2015) addresses this problem by defining a set of generative actions that maps words in the sentence to their AMR concepts and use a local classifier to learn these actions. Given such sparse data, making full use of contextual information is crucial to accurate concept labeling. Bidirectional LSTM has shown its success on many sequence la- beling tasks since it is able to combine contextual information from both directions and avoid man- ual feature engineering. However, it is non-trivial to formalize …

Content truncated to 500 words