Print

WG 2: Parsing MWEs

Internal organization

Leader: Yannick Parmentier (FR)
Vice-leader: Jakub Waszczuk (FR)

Members (by 8.9.2016)

Number:  64
Countries:  23
ESR/non-ESR:  34/30

Useful links

  WG2 webpage (here): http://typo.uni-konstanz.de/parseme/index.php/2-general/67-wg-2-parsing-mwes

  PARSEME wiki: http://wiki.studiumdigitale.uni-frankfurt.de/FB10_Parseme/index.php/Main_Page (includes registration procedure)

  WG2 wiki: http://wiki.studiumdigitale.uni-frankfurt.de/FB10_Parseme/index.php/WG2_Parsing_Techniques

  WG2 mailing list: http://chopin.ipipan.waw.pl/mailman/listinfo/parseme-wg2

Introduction

Deep parsing is the fundamental process aiming at the representation of the syntactic structure of phrases and sentences. In the traditional methodology this process is based on lexicons and grammars representing roughly properties of words and interactions of words and structures in sentences. Several linguistic frameworks, such as Head-driven Phrase Structure Grammar (HPSG), Lexical Functional Grammar (LFG), Tree Adjoining Grammar (TAG), Combinatory Categorial Grammar (CCG), etc., offer different structures and combining operations for building grammar rules. These already contain mechanisms for expressing properties of MWEs, which, however, need improvement in how they account for idiosyncrasies of MWEs on the one hand and their similarities to regular structures on the other hand. Thus, these mechanisms are rarely used on a large scale. Moreover, the semantic representation of MWEs remains a challenge. Finally, grammar development, like lexicon construction (see WG1), is considered relatively costly.

Another important challenge in parsing is the linguistic ambiguity. Most phrases and sentences yield several competing parses and in some particularly complex cases this number may rise to several hundreds or even thousands. Consequently, the parsing process is slow and raw parsing results are hardly usable. Seminal works have shown that providing mechanisms dedicated to MWEs, and fully integrated into parsing, improves both parsing efficiency and linguistic precision, since MWEs enable pruning spurious parse structures.

Objectives

Expected outcomes

Workplan

Year 1 (till March 2014)

 Listing of grammatical resources of MWEs (national projects, links, descriptions, tools, ...)

 Survey of on-going research related to parsing of MWEs within WG2 members (see talks given during the 2nd WG2 meeting in March 2014 in Athens)

Year 2 (till March 2015)

 Working subgroups dedicated to specific aspects of MWE representation and parsing (see minutes of the 3rd WG2 meeting in September 2014 in Frankfurt-am-Main)

 Following the discussions held in Frankfurt, a wiki has been setup for WG2 members to share information related to Parsing MWEs. This wiki is accessible via the following link, please register and contribute.

Year 3 (till March 2016)

WG meetings dedicated to the introduction of tools for representing and handling MWEs (see minutes of the 5th WG2 meeting in September 2015 in Iaşi)

Year 4 (till March 2017)

Collaborative book under construction to be published by Language Science Press (expected release date : March 2017).

STSMs related to WG 2