Multilingual discriminative lexicalized phrase structure parsing

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1847–1856, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Multilingual discriminative lexicalized phrase structure parsing Benoit Crabb´e Alpage – Universit´e Paris Diderot – Inria – IUF Place Paul Ricoeur 75013 Paris benoit.crabbe@univ-paris-diderot.fr Abstract We provide a generalization of discrimina- tive lexicalized shift reduce parsing tech- niques for phrase structure grammar to a wide range of morphologically rich lan- guages. The model is efficient and outper- forms recent strong baselines on almost all languages considered. It takes advantage of a dependency based modelling of mor- phology and a shallow modelling of con- stituency boundaries. 1 Introduction Lexicalized phrase structure parsing techniques were first introduced by Charniak (2000) and Collins (2003) as generative probabilistic models. Nowadays most statistical models used in natu- ral language processing are discriminative: dis- criminative models provide more flexibility for modelling a large number of variables and conve- niently expressing their interactions. This trend is particularly striking if we consider the literature in dependency parsing. Most state of the art multi- lingual parsers are actually weighted by discrimi- native models (Nivre and Scholz, 2004; McDon- ald et al., 2005; Fern´andez-Gonz´alez and Martins, 2015). With respect to multilingual phrase structure parsing, the situation is quite different. Most parsers focus on fixed word order languages like English or Chinese as exemplified by Zhu et al. (2013). Despite a few exceptions (Collins et al., 1999), multilingual state of the art results are gen- erally derived from the generative model of Petrov et al. (2006). Although more recently Hall et al. (2014) introduced a conditional random field parser that clearly improved the state of the art in the multilingual setting. Both Petrov et al. (2006) and Hall et al. (2014) frame their parsing model to model in priority regular surfacic patterns and word order: Petrov et al. (2006) crucially infers category refinements (called category ‘splits‘) in order to specialize the grammar on recurrent informative patterns ob- served on input spans. Hall et al. (2014) re- lies on a similar intuition : the model essentially aims to capture regularities on the spans of con- stituents and their immediate neighbourhood, fol- lowing earlier intuitions of Klein and Manning (2004). This modelling strategy has two main mo- tivations. First it reduces the burden of feature en- gineering, making it easier to generalize to multi- ple languages. Second it avoids modeling explic- itly bilexical dependencies for which parameters are notoriously hard to estimate from small data sets such as existing treebanks. On the other hand this strategy becomes less in- tuitive when it comes to modeling free word or- der languages where word order and constituency should in principle be less informative. As such, the good results reported by Hall et al. (2014) are surprising. It suggests that word order and constituency might be more relevant than often thought for modelling free word order languages. Nevertheless, free word order languages also tend to be morphologically rich languages. This paper shows that …

Content truncated to 500 words