Minutes from the 3rd PARSEME WG2 Meeting

Frankfurt-am-Main, 9 September 2014
Minuted by Yannick Parmentier, with the help of Timm Lichte and Stella Markantonatou

As announced on the WG2 mailing list on August 25th, the 3rd WG2 meeting was organised in a different way as the previous ones (which were consisting in scientific talks followed by questions). For the first time, the WG2 meeting took the shape of parallel sessions focussing on specific topics related to parsing MWEs, followed by a common session to summarise the activities of each session (this organisation was inspired by Action IC1005).

Parallel sessions

The topics were:

 (a) Which properties in MWEs are easy to express in grammars and which are not (challenging examples)?
     Chair: Stella Markantonatou

 (b) Can we perform compositional semantics for MWEs ?
     Chair: Timm Lichte

 (c) What kind of methods can be used to handle MWEs in pre/post-processing wrt. parsing?
     Chair: Yannick Parmentier

Each sub-session was given 1 hour 45, and the final common session 1 hour.

The minutes of the parallel sessions are available below.

Common session

During the final common session, on top of the summaries by chairs, a round table took place, in particular to discuss the future of WG2 (e.g. organisation of the next meeting in March 2015 in Malta, activity of the WG between meetings). Its outcomes are the following:

* Giuseppe Attardi proposed the following:
   1. a call for interesting issues/challenges is sent out on the WG2 mailing list in advance ;
   2. people are informed about these issues, and can apply for participating in a group addressing this issue (addressing has to be understand as a general term, any contribution is welcome, such as giving references, applying for giving a short talk about this issue at the next meeting, etc.).

   For instance, such a challenge could be: MWE-aware information extraction from parsing outputs.

   From the round table, and the discussions of the parallel sessions, the following topics have been mentioned as other potential challenges:
    + relations of MWEs with psycholinguistics
    + semantic representation of MWE

* The creation of a wiki to share experience and knowledge has been mentioned. It would give a facility for people to contribute to WG2. Such a WG2 space on a PARSEME wiki has just been set up (thanks to Manfred Sailer): http://wiki.studiumdigitale.uni-frankfurt.de/FB10_Parseme/index.php/Main_Page. A first draft of structure/content will be added soon. A call for contribution will then be sent on the mailing list. Once PARSEME members contribute to the wiki, we hope to have a clearer view on the hot topics for WG2 (so far, it seems that semantics is going to have a major role in the WG2 activities).

Future of WG2

The next steps are the following:

 + late October: call for challenges/hot topics sent out on the WG2 mailing list
 + late November: list of challenges sent out and call for interests
 + early February: definition of parallel sessions for 4th WG2 meeting, each focussing on a given challenge ; list of "assignments" for WG2 members.

The idea being that WG2 members benefit from each other's experience in tasks related to parsing MWEs.

Summaries of the parallel sessions

(a) Representation of MWEs within linguistic resources (and limits)

  Link to the summary of subgroup 2.a by Stella Markantonatou and Agata Savary.

(b) MWEs and compositional semantics

  Link to the summary of subgroup 2.b by Timm Lichte.

(c) Parsing MWEs using pre/post processing

  Link to the summary of subgroup 2.c by Yannick Parmentier.