LAW-MWE-CxG business meeting Santa Fe, 26 August 2018 Agenda - - PowerPoint PPT Presentation

LAW-MWE-CxG business meeting

Santa Fe, 26 August 2018

Agenda

• Feedback from the joint workshop
• Announcements
• The future of the PARSEME corpus and shared task
• SIGLEX-MWE section
• MWE research questions

Feedback from bringing 3 communities together

• 3 communities

○ LAW workshop - organised by SIGANN since 2007 (12th edition) ○ MWE workshop - organised by SIGLEX since 2003 (14th edition) ○ CxG - 4 past events on computational approaches to constructions

• LAW-MWE-CxG:

○ Research track: ■ 41 submissions: 22 LAW, 21 MWE, 6 CxG, 8 MWE shared task ■ 22 selected papers (16 long, 6 short): 11 cover at least 2 topics (out of the 3) ■ 65% selectivity rate ○ Shared task track: ■ 8 system description papers were submitted, all selected

Feedback from participants

• Added value from bringing 3 communities together
• Future research directions
• How to further develop synergies?

Announcements

• Phraseology and Multiword Expressions - book series at Language Science

Press, Berlin

○ fully Open Access, collaborative proofreading ○ just published: ■ Manfred Sailer, Stella Markantonatou (eds.) Multiword expressions: Insights from a multi-lingual perspective ○ to appear in 2018 ■ Yannick Parmentier, Jakub Waszczuk (eds.) Representation and parsing of multiword expressions: Current trends ■ Stella Markantonatou, Carlos Ramisch, Agata Savary, Veronika Vincze (eds.) Multiword expressions at length and in depth: Extended papers from the MWE 2017 workshop ○ 2 other books in the pipeline ○ new book proposals are welcome

Announcements

• Other announcements from the audience

PARSEME corpus

• PARSEME corpus edition 1.1

○ 20 languages, 6 mln tokens, 80,000 VMWE annotations ○

• penly available on CLARIN/LINDAT: http://hdl.handle.net/11372/LRT-2842
• Future annotation campaigns

○ New MWE categories: ■ Nominal MWEs:

• non-compositional NPs (hot dog),
• MW named entities (Red Sea),
• complex terms (recurrent neural network)

■ Adjectival MWEs: crystal clear, as busy as a bee ○ New languages (call for language leaders) ○ First step: annotation guidelines (call for contributors)

• Organize continuous corpus enhancements (regular releases)

PARSEME shared task

• Feedback from the ST edition 1.1

○

12 teams, 17 systems, 19 languages covered

○

Feedback on the .cupt format ○ Feedback on possible enhancements of the ST organisation

• ST edition 2.0:

○

2020 or later

○

Including nominal & adjectival MWEs

○

Maybe a joint task with parsing or NE recognition

UD-PARSEME coordination

• Towards the .conllup format

○

Stand-off metaformat for adding new columns to .conllu

○

The .cupt format is an instance of this metaformat for PARSEME

• MWE working group at UDs - objectives:

○ Separate syntactic and semantic annotation ○ Annotate properly the MWEs which have a correct syntactic structure ○ Solutions to the representation of syntactically irregular structures ○ Richer guidelines for UD annotation of MWEs ○ Avoiding competing semantically-driven annotations between UDs and PARSEME (e.g. LVCs or VCPs in today's UD tagset)

SIGLEX-MWE section

• MWE section of SIGLEX now has a constitution and a standing committee
• Nominations to the SC for 2018-2020

○ 1 elected representative (3-year mandate): Agata Savary ○ 4 nominated officers (2-year mandate): Francis Bond, Styliani Markantonatou, Johanna Monti, Carlos Ramisch ○ 2 officers stepping down and 2 joining every year ○ call for candidates, also open to members not present at the workshop ○ Deadlines: ■ Expressions of interest: 30 August ■ Beginning of mandate: end September

• MWE 2019 workshop

○

Co-organisation with the WordNet community

○

Preferred venue: ACL 2019 in Florence

MWE research questions

• Motivations

○ The CL/NLP community is becoming increasingly engineering-oriented. ○ It is often hard to understand the underlying research issues, the theoretical hypotheses which the experimental science is trying to (in)validate. ○ See also Joakim Nivre's ACL 2017 presidential address (fast science vs. slow science)

• Aim: better formulate the research questions and hypotheses underlying

the activities of the MWE community - see a draft

• Objectives

○ Better understanding of the state-of-the-art and perspectives of the MWE research ○ Make the MWE research more interesting ○ Lead the efforts of the community towards important challenges to be addressed ○ Pave the way towards convergences with other communities