Content Gazette - July 2014 - 19

CALL	
  FOR	
  PAPERS	
  
IEEE/ACM	
  Transactions	
  on	
  Audio,	
  Speech	
  and	
  Language	
  Processing	
  --‐	
  Special	
  issue	
  on	
  continuous--‐
space	
  and	
  related	
  methods	
  in	
  natural	
  language	
  processing	
  
Natural	
   Language	
   Processing	
   (NLP)	
   aims	
   to	
   analyze,	
   understand,	
   and	
   generate	
   languages	
   that	
  
humans	
   use	
   naturally.	
   	
   Significant	
   progress	
   in	
   NLP	
   has	
   been	
   achieved	
   in	
   recent	
   years,	
   addressing	
  
important	
   and	
   practical	
   real--‐world	
   problems,	
   enabling	
   mass	
   deployment	
   of	
   large--‐scale	
   systems.	
  	
  
New	
   machine	
   learning	
   paradigms	
   such	
   as	
   deep	
   learning	
   and	
   continuous	
   space	
   methods	
   have	
  
contributed	
   to	
   inferring	
   language	
   patterns	
   from	
   increasingly	
   large	
   real--‐world	
   data	
   and	
   to	
   making	
  
predictions	
  about	
  new	
  data	
  more	
  accurate.	
  	
  
One	
  of	
  the	
  challenges	
  in	
  NLP	
  is	
  to	
  represent	
  language	
  in	
  a	
  form	
  that	
  can	
  be	
  processed	
  effectively	
  by	
  
computing	
  algorithms.	
  Words	
  in	
  sequence	
  are	
  traditionally	
  treated	
  as	
  discrete	
  symbols,	
  which	
  has	
  its	
  
advantages	
  and	
  limitations.	
  	
  The	
  research	
  on	
  continuous	
  space	
  methods	
  provides	
  a	
  promising	
  
alternative	
  that	
  describes	
  words	
  and	
  their	
  semantic	
  and	
  syntactic	
  relationships	
  in	
  a	
  different	
  way.	
  In	
  
continuous	
  space	
  language	
  modeling,	
  we	
  represent	
  words	
  with	
  real--‐valued	
  vectors.	
  In	
  this	
  way,	
  
conditional	
  probability	
  distributions	
  of	
  words	
  can	
  be	
  learned	
  and	
  expressed	
  as	
  smooth	
  functions	
  of	
  
these	
  vectors;	
  similar	
  words	
  are	
  therefore	
  described	
  as	
  neighbors	
  in	
  a	
  continuous	
  space.	
  	
  A	
  Neural	
  
Network	
  Language	
  Model	
  is	
  a	
  typical	
  example	
  of	
  such	
  continuous	
  space	
  methods.	
  
	
  
Building	
  on	
  the	
  success	
  of	
  acoustic	
  and	
  statistical	
  language	
  modeling,	
  research	
  on	
  artificial	
  (deep)	
  
neural	
  networks	
  and	
  continuous	
  space	
  models	
  in	
  general	
  has	
  seen	
  significant	
  progress	
  in	
  mitigating	
  
data	
  sparseness,	
  incorporating	
  longer	
  contexts,	
  and	
  modeling	
  morphological,	
  syntactic	
  and	
  semantic	
  
relationships	
  across	
  words.	
  As	
  a	
  result,	
  continuous	
  space	
  models	
  are	
  now	
  embedded	
  in	
  many	
  state--‐
of--‐the--‐art	
  speech	
  recognition	
  and	
  machine	
  translation	
  systems.	
  This	
  special	
  issue	
  provides	
  a	
  forum	
  
to	
  discuss	
  the	
  latest	
  findings	
  on	
  research	
  problems	
  related	
  to	
  the	
  application	
  of	
  continuous	
  space	
  
and	
  related	
  models	
  in	
  NLP.	
  We	
  invite	
  papers	
  on	
  various	
  NLP	
  topics,	
  including	
  but	
  not	
  limited	
  to	
  	
  
	
  
*	
  Automatic	
  speech	
  recognition	
  
*	
  Text	
  mining	
  
*	
  Speaker	
  recognition	
  
*	
  Computational	
  semantics	
  
*	
  Language	
  modeling	
  
*	
  Morphological	
  analysis	
  
*	
  Machine	
  translation	
  
*	
  Syntactic	
  parsing	
  
*	
  Spoken	
  language	
  understanding	
  
*	
  Discourse	
  and	
  dialogue	
  
*	
  Spoken	
  document	
  retrieval	
  	
  
*	
  Machine	
  learning	
  methods	
  
The	
  authors	
  are	
  required	
  to	
  follow	
  the	
  Author's	
  Guide	
  for	
  manuscript	
  submission	
  to	
  the	
  IEEE	
  /ACM	
  
Transactions	
  on	
  Audio,	
  Speech,	
  and	
  Language	
  Processing	
  at	
  
http://www.signalprocessingsociety.org/publications/periodicals/taslp/taslp--‐author--‐information	
  

	
  

*	
  Submission	
  deadline:	
  Jul.	
  15,	
  2014	
  
*	
  Notification	
  of	
  first	
  round	
  review:	
  Oct.	
  1,	
  2014	
  
*	
  Notification	
  of	
  acceptance:	
  Dec.	
  1,	
  2014	
  
*	
  Final	
  manuscripts	
  due:	
  Dec.	
  15,	
  2015	
  
	
  
For	
  further	
  information,	
  please	
  contact	
  the	
  guest	
  editors,	
  
	
  
*	
  Haizhou	
  Li,	
  	
  Institute	
  for	
  Infocomm	
  Research,	
  Singapore	
  
*	
  Marcello	
  Federico,	
  FBK,	
  Italy	
  
*	
  Xiaodong	
  He,	
  Microsoft	
  Research,	
  USA	
  
*	
  Helen	
  Meng,	
  The	
  Chinese	
  University	
  of	
  Hong	
  Kong,	
  China	
  
*	
  Isabel	
  Trancoso,	
  INESC--‐ID	
  	
  and	
  Instituto	
  Superior	
  Técnico,	
  University	
  of	
  Lisbon,	
  Portugal	
  

www.signalprocessingsociety.org

[19]

JULY 2014
http://www.signalprocessingsociety.org/publications/periodicals/taslp/taslp--‐author--‐information http://www.signalprocessingsociety.org
Table of Contents for the Digital Edition of Content Gazette - July 2014
