Surface Realisation Using Full Delexicalisation

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 3086–3096, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 3086 Surface Realisation Using Full Delexicalisation Anastasia Shimorina LORIA / Lorraine University anastasia.shimorina@loria.fr Claire Gardent LORIA / CNRS claire.gardent@loria.fr Abstract Surface realisation (SR) maps a meaning rep- resentation to a sentence and can be viewed as consisting of three subtasks: word ordering, morphological inflection and contraction gen- eration (e.g., clitic attachment in Portuguese or elision in French). We propose a modular approach to surface realisation which models each of these components separately, and eval- uate our approach on the 10 languages covered by the SR’18 Surface Realisation Shared Task shallow track. We provide a detailed evalua- tion of how word order, morphological realisa- tion and contractions are handled by the model and an analysis of the differences in word or- dering performance across languages. 1 Introduction Surface realisation maps a meaning representation to a sentence. In data-to-text generation, it is part of a complex process aiming to select, compress and structure the input data into a text. In text- to-text generation, it can be used as a mean to rephrase part or all of the input content. For in- stance, Takase et al. (2016) used surface realisa- tion to generate a summary based on the meaning representations of multiple input documents and Liao et al. (2018) to improve neural machine trans- lation. By providing parallel data of sentences and their meaning representation, the SR’18 Surface Reali- sation shared task (Mille et al., 2018) allows for a detailed evaluation and comparison of surface re- alisation models. Moreover, as it provides training and test data for multiple languages, it also allows for an analysis of how well these models handle languages with different morphological and topo- logical properties. The SR’18 shared task includes two tracks: a shallow track where the input is an unordered, lemmatised dependency tree and a deep track where function words are removed and syntactic relations are replaced with semantic ones. In this paper, we focus on the shallow track of the SR’18 Shared Task and we propose a neural approach which decomposes surface realisation into three subtasks: word ordering, morphological inflection and contraction generation (e.g., clitic attachment in Portuguese or elision in French). We provide a detailed analysis of how each of these phenomena (word order, morphological realisation and con- traction) is handled by the model, and we discuss the differences between languages. For reproducibility, all our experiments includ- ing data and scripts are available at https:// gitlab.com/shimorina/emnlp-2019. 2 Related Work Early approaches for surface realisation adopted statistical methods, including both pipelined (Bohnet et al., 2010) and joint (Song et al., 2014; Puduppully et al., 2017) architecture for word or- dering and morphological generation. Multilingual SR’18 was preceded by the SR’11 surface realisation task for the English language only (Belz et al., 2011). The submitted sys- tems in 2011 had grammar-based and statisti- …

Content truncated to 500 words