Strategies for Contiguous Multiword Expression Analysis and Dependency Parsing

Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics, pages 743–753, Baltimore, Maryland, USA, June 23-25 2014. c⃝2014 Association for Computational Linguistics Strategies for Contiguous Multiword Expression Analysis and Dependency Parsing Marie Candito Alpage Paris Diderot Univ INRIA marie.candito@ linguist.univ-paris-diderot.fr Matthieu Constant Universit´e Paris-Est LIGM CNRS Matthieu.Constant@ u-pem.fr Abstract In this paper, we investigate various strate- gies to predict both syntactic dependency parsing and contiguous multiword expres- sion (MWE) recognition, testing them on the dependency version of French Tree- bank (Abeill´e and Barrier, 2004), as in- stantiated in the SPMRL Shared Task (Seddah et al., 2013). Our work focuses on using an alternative representation of syntactically regular MWEs, which cap- tures their syntactic internal structure. We obtain a system with comparable perfor- mance to that of previous works on this dataset, but which predicts both syntactic dependencies and the internal structure of MWEs. This can be useful for capturing the various degrees of semantic composi- tionality of MWEs. 1 Introduction A real-life parsing system should comprise the recognition of multi-word expressions (MWEs1), first because downstream semantic-oriented ap- plications need some marking in order to dis- tinguish between regular semantic composition and the typical semantic non-compositionality of MWEs. Second, MWE information, is intuitively supposed to help parsing. That intuition is confirmed in a classical but non-realistic setting in which gold MWEs are pre- grouped (Arun and Keller, 2005; Nivre and Nils- son, 2004; Eryiˇgit et al., 2011). But the situation is much less clear when switching to automatic MWE prediction. While Cafferkey et al. (2007) report a small improvement on the pure parsing 1Multiword expressions can be roughly defined as con- tinuous or discontinuous sets of tokens, which either do not exhibit full freedom in lexical selection or whose meaning is not fully compositional. We focus in this paper on contiguous multiword expressions, also known as “words with spaces”. task when using external MWE lexicons to help English parsing, Constant et al. (2012) report re- sults on the joint MWE recognition and parsing task, in which errors in MWE recognition allevi- ate their positive effect on parsing performance. While the realistic scenario of syntactic pars- ing with automatic MWE recognition (either done jointly or in a pipeline) has already been investi- gated in constituency parsing (Green et al., 2011; Constant et al., 2012; Green et al., 2013), the French dataset of the SPMRL 2013 Shared Task (Seddah et al., 2013) only recently provided the opportunity to evaluate this scenario within the framework of dependency syntax.2 In such a sce- nario, a system predicts dependency trees with marked groupings of tokens into MWEs. The trees show syntactic dependencies between se- mantically sound units (made of one or several tokens), and are thus particularly appealing for downstream semantic-oriented applications, as de- pendency trees are considered to be closer to predicate-argument structures. In this paper, we investigate various strate- gies for predicting from a tokenized sentence both MWEs and syntactic dependencies, using the French dataset of the SPMRL 13 Shared Task. …

Content truncated to 500 words