Deep Lexical Segmentation and Syntactic Parsing in the Easy-First Dependency Framework

Proceedings of NAACL-HLT 2016, pages 1095–1101, San Diego, California, June 12-17, 2016. c⃝2016 Association for Computational Linguistics Deep Lexical Segmentation and Syntactic Parsing in the Easy-First Dependency Framework Matthieu Constant♠♦ Joseph Le Roux♣ Nadi Tomeh♣ ♠ Universit´e Paris-Est, LIGM, Champs-sur-Marne, France ♦ Alpage, INRIA, Universit´e Paris Diderot, Paris, France ♣ LIPN, Universit´e Paris Nord, CNRS UMR 7030, Villetaneuse, France matthieu.constant@u-pem.fr, leroux@lipn.fr, tomeh@lipn.fr Abstract We explore the consequences of representing token segmentations as hierarchical structures (trees) for the task of Multiword Expression (MWE) recognition, in isolation or in com- bination with dependency parsing. We pro- pose a novel representation of token segmen- tation as trees on tokens, resembling depen- dency trees. Given this new representation, we present and evaluate two different archi- tectures to combine MWE recognition and de- pendency parsing in the easy-first framework: a pipeline and a joint system, both taking ad- vantage of lexical and syntactic dimensions. We experimentally validate that MWE recog- nition significantly helps syntactic parsing. 1 Introduction Lexical segmentation is a crucial task for natural language understanding as it detects semantic units of texts. One of the main difficulties comes from the identification of multiword expressions [MWE] (Sag et al., 2002), which are sequences made of mul- tiple words displaying multidimensional idiomatic- ity (Nunberg et al., 1994). Such expressions may ex- hibit syntactic freedom and varying degree of com- positionality, and many studies show the advantages of combining MWE identification with syntactic parsing (Savary et al., 2015), for both tasks (Wehrli, 2014). Indeed, MWE detection may help parsing, as it reduces the number of lexical units, and in turn parsing may help detect MWEs with syntactic free- dom (syntactic variations, discontinuity, etc.). In the dependency parsing framework, some pre- vious work incorporated MWE annotations within syntactic trees, in the form of complex subtrees ei- ther with flat structures (Nivre and Nilsson, 2004; Eryi˘git et al., 2011; Seddah et al., 2013) or deeper ones (Vincze et al., 2013; Candito and Constant, 2014). However, these representations do not cap- ture deep lexical analyses like nested MWEs. In this paper, we propose a two-dimensional repre- sentation that separates lexical and syntactic layers with two distinct dependency trees sharing the same nodes1. This representation facilitates the annota- tion of complex lexical phenomena like embedding of MWEs (e.g. I will (take a (rain check))). Given this representation, we present two easy-first depen- dency parsing systems: one based on a pipeline ar- chitecture and another as a joint parser. 2 Deep Segmentation and Dependencies This section describes a lexical representation able to handle nested MWEs, extended from Constant and Le Roux (2015) which was limited to shallow MWEs. Such a lexical analysis is particularly rele- vant to perform deep semantic analysis. A lexical unit [LU] is a subtree of the lexical seg- mentation tree composed of either a single token unit or an MWE. In case of a single token unit, the sub- tree is limited to a single node. In case of an MWE, the subtree is rooted by its …

Content truncated to 500 words