—- AIM-WEST Kickoff (skype) meeting 26th March 2014 —-

(notes by L. Besacier)
0. trying to connect…

finally workout at 4:20… 😉
1. Reminder of the project themes and goals (Carlos, Laurent) – 5min
investigate techniques, resources and protocols for evaluating and integrating models of multiword expression (MWE) processing into machine translation and automatic speech recognition technology.
Tasks (focus on 1 & 2 for the moment!)
1 Corpora construction
2 Protocols for translation evaluation
3 Pre-processing of corpora
4 Lexical and ontological resource construction for French, Portuguese and English
5 Cognitive and data intensive models of MWE learning
6 Lexical models of MWEs
7 Syntactic models of MWEs
8 Semantic models of MWEs
9 Automatic speech recognition systems
10 Machine translation
11 Reports, papers and articles
2. Interests and resources of each participant group (Helena, Aline, Thierry, Laurent, Carlos) – 5min each
Helena: ML approach (never ending learning from Internet) ; monolingual translation (paraphrasing) ; multilingual (portuguese – english) extraction –
simplify input for the translation system (possible for english) => find an equivalent ‘singe word’ expression instead of a MWE one …
An undergraduate student (Alexandre Rondon) will work on that, supervised by Helena and Carlos
Thierry:  LATTICE do not work on MT / cognitive models for MWE / Learning to detect new named entities (MWE) / Acquisition of verbal expressions / Work mainly on French and English
Aline: (described by Carlos) : MWEs, shift to semantic and distributional representations of expressions (to model semantic compositionality), other NLP subjects (text simplification, speech recognition, language acquisition models),
Laurent: see summary of team web site on
Carlos: LIF/Marseille – find bilingual MWEs (jointly) ; model the semantics of these expressions ; parsing : ASR (error detection) ; multilingual lexical resources
3. Financial details and technicalities -> letters, funded actions, scholarships, travels (Aline, Helena, Laurent) – 5min
-10keuros received on the French side – to be used by the end of 2014 !!
We will have 10k in 2015 and 10k in 2016 too…
-San Paulo : 48k reals (for the 3 years) ; they got the money
-Porto-Alegre : not yet ; they asked for 120k reals (=40keuros)
==>letters from LIG will be sent to both Helena & Aline (please, send me your precise mailing adresses)
4. Provisional schedule and budget for 2014 (Aline, Helena, Laurent) – 5min
For 2014, we will probably send confirmed researches – Oct/nov 2014 (>20th Oct.) in San Carlos is an option (?) –
Maybe we should focus more in 2014 on collaboration with San Carlos –
Helena can also go to Porto-Alegre (Aline will ask to her collegues about that)
3 missions (Fr=>Br) should be ok for 2014
Proposition: we could do San Carlos (4-5 days) + then Porto Allegre … 4/4 or 5/5 days …
Also, don’t forget to mention AIM-WEST when you publish paper on the topic …
5. Communication, website, mailing list (Laurent) – 5min
Wiki-based (wordpress) web site – LIG will handle it – LIG can handle the mailing-list too
6. Actions – shared task proposal (Carlos, Laurent) – 5min
shared task : LIG will have something (a challenging corpus for MWE Translation) at LIG for french-english…from that we may extend it to portuguese-english … we work together in oct-nov to define precisely the shared task, corpora, metrics…and then we launch it end of 2014 ! …
7. General discussion and questions – 5min
Carlos : will send a doodle for the next phone meeting (end of April) in order to schedule Oct-Noc meeting in San Carlos
(Aline will be in Marseille/Paris in May…we’ll see about a meeting at this moment…)